ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE QUENTIN WODON, CHLOE FÈVRE, CHATA MALÉ, ADA NAYIHOUBA, AND HOA NGUYEN JULY 2021 © 2021 The International Bank for Reconstruction and Development - The World Bank, Washington, DC. All rights reserved. This study was produced by a team at the World Bank with funding from the Global Partnership to End Violence Against Children. The findings, interpretations, and conclusions expressed in the study are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. The World Bank does not guarantee the accuracy of the data included in this work. Suggested citation: Wodon, Q., C. Fèvre, C. Malé, A. Nayihouba, and H. Nguyen. 2021. Ending Violence in Schools: An Investment Case. Washington, DC: The World Bank and the Global Partnership to End Violence Against Children. Rights and Permissions: The material in this work is subject to copyright. Information and illustrations contained in this report may be freely reproduced, published, or otherwise used for noncommercial purposes without permission from the World Bank. However, the World Bank requests that the original study be cited as the source. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. I | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 TABLE OF CONTENTS Executive Summary 1 Introduction 7 Chapter 6: Promising School-Based Interventions 51 Part I: Violence in School and Its Impacts 9 Lifecycle Approach 51 Chapter 1: Types of Violence and Prevalence 9 Early Childhood: Preventing Child Maltreatment 51 and Enhancing Parenting Skills Types of Violence 9 Secondary Education: Anti-bullying Interventions Prevalence of Violence in Schools 11 55 and Gender Equality Programs Continuum of Violence, Poly-victimization, Referral Pathways: What To Do When Cases of 12 58 and Vulnerable Groups Violence Are Identified Chapter 2: Impacts on Education and Earnings 15 Beyond Specific Programs: The Importance of Laws 59 Students’ Experience in School 15 Chapter 7: Engaging All Stakeholders 60 Learning in School 19 Whole School Approach 60 Reasons for Dropping Out of School 24 Enhancing Teachers’ Skills on Positive Discipline 61 Completion Rates and Educational Attainment 26 and Classroom Management Indirect Effects through Educational Attainment 29 Engaging with Parents and Caregivers 61 Chapter 3: Multi-pronged Negative Impacts on Working with Communities on Social Norms and 32 65 Children Beyond Education and Earnings Safe Passage To Schools Injuries 32 Strategic Approaches for Violence Prevention 65 Risky Behaviors and Student Perceptions of Chapter 8: Illustrative Cost-Benefit Analyses 68 33 Health and Well-being Principles and Limits of Cost-Benefit Analysis 68 Indirect Effects through Educational Attainment 38 Early Childhood Interventions 69 Chapter 4: Cost of Inaction 40 Socio-Emotional Skills and Related Programs 71 Literature on the Cost of Violence against Children 40 Anti-Bullying Interventions 72 Potential Losses in Human Capital Wealth from 44 Reducing Unit Costs and Scaling Up 73 Violence in Schools Part II: Programs and Policies 47 References 74 Chapter 5: Factors Leading to Violence 47 Risk Factors 47 Protective Factors 49 JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | II ACKNOWLEDGMENTS This study was produced by a team at the World Bank with funding from the Global Partnership to End Violence Against Children. The authors are very grateful to Rafael Hermoso, Matthew De Galan, Marija Manojlovic, Emma Green, Katharina Anton-Erxleben, and Howard Taylor at the Partnership for their support for this work and multiple comments and suggestions at various stages of the work. The team benefitted from advise and comments from a partners’ advisory group including (by institutional alphabetical order) James Mercy (Center for Disease Control), Elena Ahmed and Martin Niblett (DFID), Xiangming Fang (Georgia State University), Leila Asrari and Debora Cobar (Plan International), Bill Bell (Save the children UK), Sally Beadle and Christophe Cornu (UNESCO), Sujata Bordoloi (UNGEI), Stephen Blight, Rocio Aznar Daban, and Clarice Da Silva e Paula (UNICEF), Andrew Claypole (United Nations), and Elena Gaia and Tamara Tutnjevic (World Vision International). At the World Bank, the authors benefited from overall guidance from Jaime Saavedra, Omar Arias, and Meskerem Mulatu and comments and suggestions among others from Diana J. Arango, Deon Filmer, Kurt Larsen, Fernando Lavendez, Laura McDonald, Manal Quota, and Eliana Rubiano. Support for communications was provided by Andrew Hassett at the Global Partnership to End Violence Against Children and by Karolina Ordon and Kristyn Schrader-King at the World Bank. The authors are also grateful to the team at Together for Girls for granting access to selected Violence against Children surveys and to the team at Young Lives team for access to their surveys. The opinions expressed in the report are those of the authors only and need not reflect the views of the World Bank, its Executive Directors, of the countries they represent. Any errors or omissions are only those of the authors. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | III EXECUTIVE SUMMARY victimization, physical and psychological exploitation, cyber victimization, bullying5 , fights, and sexual violence. It also Ending violence in and around school (VIAS) is essential includes violence by teachers such as corporal punishment, to reap the benefits from education and ensure children’s with potential negative impacts6 . Overall, VIAS has major well-being. Receiving an education of good quality is the effects on children’s well-being and health, and through right of every child, as enshrined in the Convention on reduced attainment and achievement, it reduces earnings the Rights of the Child. Education plays a unique role in and productivity in adulthood. promoting respect for human rights, and contributing to safe and inclusive societies that do not condone the use Violence remains ubiquitous in schools. Table ES1 provides of violence, but rather provide children with the skills they a summary of estimates of violence in schools from survey will need as adults to find peaceful solutions to conflicts. data from multiple countries. According to these data, more Education also plays a fundamental role in countries’ ability than one in three children are subjected to physical violence, to achieve the targets set forth under the Sustainable and almost one in three are bullied. The prevalence of sexual Development Goals1 . But in addition, in today’s fast changing violence is much lower, but likely underestimated especially world, education is the foundation of countries’ future when the topic remains taboo. Girls and boys experience economic development. It drives human capital wealth (the VIAS in different ways. Apart from differences in exposure value of the future earnings of the labor force), which itself to sexual violence, girls are more likely to experience verbal accounts for two thirds of the changing wealth of nations2. and emotional abuse whereas boys are more likely to be Unfortunately, VIAS remains widespread in developing physically abused. Unfortunately, data on sexual harassment and developed countries alike, preventing children from which affects mostly girls are not widely available. VIAS is learning and leading them to drop out of school3 . Failing to an issue in virtually all countries of the world even if across prevent VIAS will affect not only children today, but also the regions, sub-Saharan Africa , South Asia, and the Middle members of their future families, their communities, and East and North Africa tend to have higher prevalence rates. societies as a whole. Due to the COVID-19 pandemic, over Globally, VIAS is likely to affect well above a half billion the last year schools were closed for substantial periods of children each year (including those in primary schools). time in many countries, but several of the factors that may lead to higher violence against children overall and violence on schools in particular have been exacerbated. The need to end violence in school is even more pressing today. HIGH PREVALENCE OF VIOLENCE IN SCHOOLS VIAS is a threat to both schooling and learning, as well as to children’s well-being, health, and future earnings as adults. The World Health Organization (WHO) has defined violence as “the intentional use of physical force or power, threatened or actual, against a person or group that results in or has a high likelihood of resulting in injury, death, psychological harm, maldevelopment, or deprivation.”4 Violence is the result of an abusive use of force. The harm can be actual or threatened. It can lead to injury or death, but also to trauma or other mental health symptoms. Violence is often multidimensional, as individuals are often subjected to multiple forms of violence and in multiple locations. VIAS includes but is not limited to child 1 Wodon et al. (2018). 2 Lange et al. (2018). 3 UNICEF (2018). See also UNESCO (2019). On violence against children more generally, see Hillis et al. (2016). See also UNICEF (2014), Office of the SRSG on Violence against Children (2016), Hillis et al. (2016), UNICEF (2017, 2019), Know Violence in Childhood (2017), and UNESCO (2019). 4 Krug et al. (2002), based on WHO Global Consultation on Violence and Health (1996). 5 Bullying is defined as repeated aggression (physical, verbal or psychological) over a prolonged period of time among peers who have an imbalance of power. 6 Naz et al. (2011). 1 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Table ES.1: Prevalence of Violence in Schools in Multi-country Surveys (%) GSHS HBSC PISA VACS DHS GSHS, HBSC, and PISA surveys Attacked in last 12 months 37.8 - - - - Involved in fight in last 12 months 27.6 31.2 - - - Injured in last 12 months 31.3 44.5 - - - Injured from fight 1.5 1.8 - - - Bullied in last 30 days 29.5 29.0 - - - Others left me out of things - - 36.8 - - Students made fun of me - - 42.7 - - Threatened by other students - - 23.8 - - Others destroyed my things - - 26.5 - - Hit by other students - - 23.4 - - Nasty rumors about me - - 33.5 - - VACS and DHS surveys - - - - Physical violence in schools - - 28.7 - Emotional violence in schools - - 3.5(*) - Sexual violence in schools - - 2.8(*) 1.5(*) Source: Authors’ estimates. Note: (*) The prevalence of emotional and sexual violence in schools in VACSs and DHSs may be underestimated. Corporal punishment by and corruption among teachers account to respond to multiple layers of risks for children and school officials (which may involve threats of violence and target the most vulnerable children9. against students) also remain widespread. Some 67 countries still do not have legislation banning corporal The COVID-19 pandemic is likely to have exacerbated punishment in schools7. In Francophone Africa, data on some of the factors that lead to violence against children corporal punishment from PASEC suggest that more in general, and violence in schools in particular. Many than a third of teachers in sixth grade of primary school individuals and households are under stress due among use corporal punishment in the classroom, leading to others to social isolation, losses in employment and income, almost two-thirds of students being beaten by teachers. and illnesses or death from the pandemic. Initial predictions Afrobarometer data for three dozen African countries of the economic impacts of the crisis were dire10 for both suggest that one in five individuals are affected by petty developed11 and developing countries12. Over time many corruption, which may lead to bribes, gifts, or favors, but projections were further revised downward, and while a may also in some cases to sexual violence against children. recovery is underway, the consequences of the crisis have been severe. Estimates suggest that the crisis may lead to The various forms of violence often do not occur in an increase in the number of poor people of 150 million13 . isolation. Instead, they tend to reinforce each other. Of those, about half are children. According to the World Children are often victims of violence in separate locales, at Food Programme, the number of people suffering from school but also at home and in the community. This feeds acute hunger may have doubled14 . Student learning suffers into a self-reproducing cycle8 or poly-victimization, which during recessions15 , and estimates suggest that learning has negative multiplier effects on children’s wellbeing and poverty – the share of children not able to read and capacity to learn, leading to higher risk of lasting physical, understand a simple text, may increase by 10 percentage mental and emotional harm. Policymakers and stakeholders points under a pessimistic scenario16 . working in schools must take poly-victimization into 7 See https://endcorporalpunishment.org/schools/. 8 Wilkins et al. (2018). 9 Finkelhor et al. (2011). 10 International Monetary Fund (2020). 11 For Europe, see European Commission (2020). 12 For sub-Saharan Africa, see World Bank (2020a). 13 World Bank (2020`b). 14 Food Security Information Network (2020). School lunch programs were also affected. These programs serve many children (World Food Programme, 2013, 2020). 15 Shores and Steinberg (2019). 16 Azevedo (2020). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 2 For schooling, girls are especially likely to be affected, the socio-economic background of the student, a leading to higher risks of child marriage18 with major disability, or factors such as teacher absenteeism, the implications for the rest of their life19. More generally, level of education of teachers, or selected characteristics children from vulnerable groups, including not only girls and of schools. Accounting for the particular level of those in extreme poverty, but also children with disabilities, violence in school observed in each country and the refugees and internally displaced persons, and other groups share of human capital wealth attributed to education are especially at risk 20 . in each country, human capital wealth could have been be higher by US$ 5.7 trillion without violence in school. LARGE ECONOMIC COSTS OF VIOLENCE IN SCHOOLS • Children dropping out of school: In a few countries where household surveys include VIAS as one of the The cost of VIAS in lost future lifetime earnings could possible reasons for dropping out of school, VIAS be of the order of US$11 trillion. In absolute value, this accounts typically for part of drop-outs at the primary estimate is slightly higher than other estimates available in or secondary levels. Based on available data for a few the literature, but it corresponds to a lifetime cost while most countries, the assumption is that across all countries, five estimates in the literature are annual cost. The estimate of percent of drop-outs may be due to VIAS. If such drop- US$11 trillion in lost human capital wealth represents slightly outs could have been avoided, and if children would have more than one percent of the baseline level of human capital completed their secondary education, human capital wealth globally. The estimate consists of two separate values: wealth could have been higher by US$ 3.9 trillion. • Children learning less in school: Based on regression The combined loss from the potential impact of VIAS analysis using data from the PISA and PASEC on educational achievement and attainment is valued at international student assessments, ending violence US$ 9.6 trillion in 2014. Given population growth and in school could result in aggregate gains in learning gains in standards of living between 2014 and 2019, that of about two percent versus baseline values21 . These value is scaled up by the ratio of global GDP in current aggregate effects may seem limited, but at the margin, US$ in both years22, leading to an estimated loss in human the effects for students affected by violence are often capital wealth in 2019 (pre-COVID) of just under US$ 11 larger than the potential impact of variables capturing trillion. This estimate is only an order of magnitude and a Table ES.2: Potential Loss in Human Capital Wealth Due to Violence in School Total Cost (US$ trillions) Human capital wealth in 2014 736.9 Losses due to children not learning as much Loss in human capital wealth for countries included in PISA 4.7 Scaled-up losses for all countries with measures of human capital wealth 5.7 Losses due to children dropping out of school Potential gain in human capital wealth with universal secondary education 77.3 Share of drop-outs due to violence in school 5% Potential gain from ending school drop-outs due to violence in school 3.9 Combined estimate of losses Combined losses from lack of learning and children dropping out of school 9.6 Potential gain from ending violence in school as share of human capital wealth 1.3% Scaling-up factor from 2014 to 2019 Cost of VIAS in 2019 (using ratio of global GDP between 2019 and 2014 at 1.1) 10.6 Source: Authors’ estimates. 17 See UNDP (2015), Onyango et al. (2019), and Bandiera et al. (2019). See also World Bank (2020g) for a review, as well as Asfaw (2018) on Ethiopia, Dureya et al. (2007) and Cerutti et al. (2019) on Brazil, and Lim (2000) on the Philippines. 18 Wodon et al. (2016, 2017); Kassa et al (2019). 19 Wodon et al. (2018). 20 World Bank (2020c). 21 Our findings confirm previous results from the literature based on country-level analysis and qualitative studies suggesting a negative impact of violence in schools in learning apart from its impact on schooling. 22 GDP per capita and human capital wealth are highly correlated. 3 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 range of potential costs could be provided using different are observed for suicidal thoughts and early childbearing assumptions. The actual costs for the two impacts included (having a child before 18). in Table ES.2 could be lower or higher. Taking into account the population of students enrolled in These estimates are approximations only suggesting orders secondary schools in the various countries, ending violence of magnitudes, but they likely represent lower bounds. in school would have beneficial effects for tens of millions These estimates are nothing more than broad orders of of students. Simulations suggest that ending violence in magnitude – they are not meant to be precise or definitive secondary schools could reduce the number of students given the many assumptions involved. But they appear to who (typically over the past 12 months) (1) have difficulties be of a reasonable order of magnitude in percentage terms sleeping by 24 million; (2) miss school by 20 million; and from base values when compared to more detailed country (3) consider suicide by 15 million. Many of these effects level estimates of losses in earnings due to VIAS. However, could be observed for children in primary school as well. In the estimates are likely to be a lower bound since they do that case, the number of children affected could be up to not factor other costs of VIAS such as health-related costs23 , two and a half times larger than those estimates. Ending nor do they account for all the benefits that would arise from violence in schools could also reduce the number of children ending VIAS in terms of gains in standards of living. in secondary schools who (1) have sex by 26 million; (2) drink alcohol by 26 million; (3) smoke by 17 million; and (4) use OTHER NEGATIVE WIDE-RANGING EFFECTS drugs by 8 million. Finally, when girls drop out of school due OF VIOLENCE IN SCHOOLS to violence in school, this increases the likelihood that they will marry as children or have early pregnancies. This can Apart from leading to losses in learning and drop-outs, have major negative consequences for the rest of their life, VIAS is highly detrimental for students’ experience in not only in terms of lower earnings in adulthood, but also for school as well as their health and well-being. If VIAS their maternal health and the health of their children, their were eliminated, this would have potentially large effects ability to make decisions within their household, and the risk for a wide range of indicators of health and well-being. of intimate partner violence. For example, for perceived health, surveys ask questions on difficulties sleeping, having headaches, stomach-ache, COMPLEX FACTORS LEADING TO VIOLENCE or back-ache and a self-assessment of health. For risky AND PROMISING INTERVENTIONS behaviors, questions are asked about whether the children have ever smoked, used alcohol, drug or cannabis, or had Ending VIAS requires multifaceted interventions, but sex. Finally, for psychological well-being, questions are promising interventions have high benefits to costs ratios. asked about whether the children ever considered suicide, Risk factors for violence include factors at the levels of the planned to commit suicide, or attempted to commit individual, the household, the community, and society. The suicide. Questions are also available on whether children are accumulation of risk factors often explains why an individual feeling low, irritable, nervous, or dizzy. In virtually all cases, behaves more violently or is more prone to be victimized experiencing VIAS is associated with worse indicators than others. Instead of looking for a single best intervention after controlling for other factors that may affect these that would be most effective in preventing violence, it indicators. Some of the largest effects are observed for the often makes sense to combine interventions that can both probabilities of feeling bad about one’s health (reduction mitigate the most salient risk factors and enhance relevant of 6.1 percentage points if VIAS were eliminated), trusting protective factors in a given context and for a specific other people (increase by 6.0 points), having suicidal group. There is no unique way to categorize programs to thoughts (reduction by 5.3 points), having sex before prevent VIAS, but a lifecycle approach is useful because the age of 18 (reduction by 4.9 points), and feeling close risk factors leading to VIAS evolve over time in a child’s life. to one’s biological father (increase by 4.9 points). In proportional terms from base values, the largest effects 23 On previous estimates at the country, regional, or global levels, see Pereznieto et al. (2010), Fang et al. (2012), Pereznieto et al. (2014), Fearon and Hoeffler (2014), Fang et al. (2015), Fang et al. (2015), Fang (2015), McCarthy et al. (2016), Fang et al. (2017), UNICEF (2019), and Deloitte Access Economics (2019). 24 School safety issues at the school level can compound each other. For example poor infrastructure and lack of basic services at schools is associated with increases in violence. Community factors, such as conflict/ fragility can also influence school level relationships and contribute to exclusion and negative behaviors such as bullying/violence. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 4 In particular: multiple strategies to develop a common vision and shared values and rules for the school, and works through the ost-benefit analyses suggest that promising • C curriculum, teacher training, parental engagement, and interventions have high benefits to costs ratios. student learning towards a safe and inclusive school climate While these ratios are sensitive to assumptions used in and respectful school values. One example is the Good the analyses, results suggest that reducing violence in School Toolkit (GST) in Uganda. Evaluations suggest that and around schools is a smart economic investment. after 18 months of implementation, the program reduced While most of the available analyses are from developed the risk of physical violence by teachers and school staff countries, programs should generate high benefits against students by 42 percent; halved the number of in developing countries as well if one presumes teachers who reported using physical violence against that results of a similar magnitude could apply. students; and improved students’ connectedness and sense of safety and belonging with their school. The program also • Early childhood interventions are essential to prevent increased teachers’ satisfaction in their role at school and VIAS and often have high returns. Cost-benefit increasing students’ wellbeing and sense of safety at school29. analyses have been implemented mostly for center- Importantly, if the GST program were implemented at scale, based interventions (typically preschools, although many unit costs for implementation would be low30. programs also include home visiting, parenting advice, health and nutrition services, and referrals for social Supporting teachers to enhance their skills in positive services)24 . Such programs tend to have high returns discipline and classroom management is also effective. on investment, with benefit to cost ratios ranging from Providing teachers with skills to improve their relationship 2.04 to 16.14. Some of the more recent programs do with students and manage behaviors lessens disruptive and not have as high benefits to cost ratios, but this may be aggressive behaviors in the classroom and enhance prosocial due in part to the fact that some of the benefits from behaviors later in life. By contrast, punitive interactions these programs in adulthood could not yet be measured. tend to feed a vicious circle of violence, delinquency, and further exclusion. The Global Initiative to End All Corporal • In primary schools, programs helping children Punishment of Children and the WHO handbook on improve their social and emotional skills also have school violence prevention31 provide useful resources on high returns. A recent synthesis of cost-benefit positive discipline for teachers and schools. When teachers analyses for these types of programs suggest and the entire school community understand that respect benefit to cost ratios ranging from 3.46 to 13.91 and trust are key pillars for child’s healthy development and across interventions in baseline scenarios25 . that corporal punishment is not only counterproductive but negatively impacts a child’s learning, the school climate can n secondary schools, a key area of focus should be to • I be transformed. reduce bullying. Reviews suggest that intensive and long- lasting programs are needed to change behaviors, with Families need to be part of school programs. Engaging with parental sessions contributing to success26 . Cost-benefit parents of adolescents that display behavioral problems can analyses have been conducted especially for the Olweus yield significant results even in a relatively short period of Bullying Prevention Program and the KiVa anti-bullying time. But parenting programs should follow evidence-based program. For the Olweus program, a benefit-cost ratio of practices, including focusing on positive discipline, positive 6.94 is suggested when start-up costs are not included27. communication, and increased bonding among family For KiVa, analyses suggest benefit-cost ratios well above members. As with teachers, providing alternative tools and one, with differences depending on countries28 . skills to caregivers in dealing with their children can help break the intergenerational cycle of violence. Effectively engaging So-called whole school approaches can help reduce with parents requires choosing wisely among alternative VIAS at a limited cost. Engaging with the entire school programs, as well as recruiting parents and keeping them community is beneficial. A whole school approach uses engaged. The most challenging part is to keep parents engaged 25 Dalziel et al. (2015). 26 Belfield et al. (2015). 27 Farrington and Ttofi (2009). 28 Highmark Foundation (2018). 29 See Huitsing et al. (2019) and McDaid (2017). 30 Devries et al. (2015). 5 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 long enough to produce sustained behavioral change, but Several guides exist in that respect, including on engaging techniques have been developed to do so. religious leaders to end VIAS. Codes of conducts and zero tolerance policies towards violence by teachers need to be Engaging with communities to shift norms also matters. adopted. More generally, four steps in the strategic process The SASA! program is a good example of how norms can be can be suggested: (1) Setting clear standards for all including challenged. SASA! means “Now!” in Kiswahili. The program through codes of conduct and appropriate laws including employs multiple strategies to build a critical mass of engaged on corporal punishment; (2) Establishing a solid diagnostic; community members, leaders, and institutions, including (3) Developing a common vision and action plan with local activism, media and advocacy, communication materials, accountability mechanisms; and (4) Promoting a whole school and training. In comparison to control communities, SASA! approach to enhance students’ connectedness with schools communities reported a reduction in levels of violence against and ensure a positive learning environment. women of 52 percent; an increase in the share of women and men who believe it is acceptable for women to refuse sex of Finally, better data are needed both to update existing 28 percent; and an increase of 50 percent in the share of school health surveys in many countries and to ensure men and women who believe that physical violence against a that broader information is collected, especially on sexual partner is unacceptable32. Essentially, SASA! works with key violence. Figures on VIAS may represent lower bound stakeholders at the community level to deconstruct power estimates, especially because data are not widely available in intimate partnerships. Another interesting program is the for some forms of violence. In some cases, prevalence may Bell Bajao! (Ring the Bell) campaign in India. Engaging with be underreported, as is likely the case for sexual violence. community is also important to ensure safe passage to schools In addition, in many countries, the available data are dated by identifying hot spots where children may feel vulnerable, and school health surveys should be updated. Improving and and placing adult monitors on those spots. expanding questionnaires in various existing surveys would also be highly valuable. These various interventions and approaches have proven benefits, but they are not exhaustive in terms of the types of programs and policies that may help prevent violence in CONCLUSION school or cope with its effects. Guidance on how to prevent violence in school is available from the WHO Handbook on Preventing VIAS is a moral imperative. It is also a smart school-based violence prevention33 and for violence against investment. The negative effects of VIAS are widespread. children more broadly from the INSPIRE framework34 . Children’s life is profoundly affected when they are victims Also relevant is the new strategy adopted by the Safe to or perpetrators of violence, with scars that last a lifetime. Learn initiative35 to which a wide range of organizations are Violence in school affects virtually all aspects of a child’s well- contributing. The organizations that are member of the Safe being, including especially mental health. Preventing VIAS is to Learn initiative have made the prevention of violence in above all a moral imperative. But it is also a good investment. schools a priority in their own strategies. For example, at the This study provides policy makers with rigorous data and World Bank, the Safe and Inclusive Schools Initiative is one evidence on both the negative impacts of VIAS and the large of five pillars of the Bank’s approach to realize the future benefits of ending VIAS. The case for investments towards of learning36 , with additional analytical work underway to ending VIAS is clear. Promising interventions to end VIAS operationalize safety in schools38 . are available and have large benefits to cost ratios. This study has been undertaken within the context of the multi-partner Beyond efforts in individual schools, strategies to end Safe to Learn global initiative to end VIAS. At the World VIAS should be led by Ministries of Education with other Bank, the study is part of a program of work on safe and Ministries or agencies. To sustainably shift norms, parent inclusive schools. The findings from the study suggest that in associations and teacher unions, as well as religious groups addition to being a moral imperative, ending VIAS is also and political parties, need to participate and be heard. a smart economic investment. 31 Greco et al. (2018). 32 World Health Organization (2019). 33 Abramsky et al. (2014). 34 WHO (2019). 35 WHO (2018). 36 Safe to learn Initiative (2021). 37 World Bank (2020d). The five pillars are: (1) learners are engaged; (2) teachers facilitate learning; (3) learning resources are adequate and diverse; (4) schools are safe and inclusive; and (5) and systems are well managed. 38 The World Bank is preparing a study on how to operationalize work in this area – both to prevent violence and consider other aspects such as the safety of school infrastructure. That forthcoming study will consider aspects of safe schools related to teaching and learning, school level relationships, safety, and the institutional environment. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 6 Unfortunately, violence against children, including VIAS, INTRODUCTION remains widespread in developing and developed countries alike and the COVID-19 pandemic is likely to have Violence against children is a violation of their human rights. exacerbated some of the factors that lead to violence against It results in significant impacts on their health and wellbeing, children in general, and violence in schools in particular. affecting in turn the enjoyment of other rights. Preventing Given its negative long-term consequences for children’s violence against children is a moral imperative. It is also a wellbeing and their capacity to learn, preventing VIAS is smart investment. This study makes the case for more and simply a prerequisite to enable children to thrive. Failure better investments towards ending violence in and around to do so will negatively affect not only the children today, schools (VIAS). It also suggests ways to achieve that goal but also the members of their future families as adults and based on the available evidence and lessons from international societies as a whole. experience. This study was undertaken within the context of the Safe to In today’s fast changing world, education – and more Learn global initiative dedicated to ending violence against generally human capital, is the foundation of countries’ future children in and around schools40. The initiative set out in a five- economic development. Education is essential to achieve point call to action to (1) Implement policy and legislation; (2) the Sustainable Development Goals (SDGs) – not only for Strengthen prevention and response at the school level; (3) SDG4 which is about education, but also for many other Shift social norms and behavior change; (4) Invest resources areas. Education also has a unique role to play in promoting effectively; and (5) Generate and use evidence. The study respect for human rights, and contributing to safe and aims to contribute to these commitments, and especially to inclusive societies that do not condone the use of violence, the generation and use of evidence on how to end violence in but rather provide children with the skills they will need as schools, and how to invest resources to do so effectively. adults to find peaceful solutions to conflicts and advance gender inequality. One of the drivers of violence against At the World Bank, the study is part of a broader program children including in schools is actually gender inequality, of work on safe and inclusive schools, which is one of the five with girls facing violence from pre-birth, with sex selective pillars of the World Bank’s strategy to improve learning41. abortions, into adulthood39. Safe and inclusive societies that The study is organized in two parts. The first part documents do not condone violence against children, but condemn it are the prevalence of VIAS, its potential impacts, and the costs societies that value girls and boys equally. associated with those impacts. Chapter 1 first defines the 39 See for example the Lancet series at https://www.thelancet.com/series/gender-equality-norms-health. 40 At the time of writing, the coalition behind Safe to Learn includes the United Nations Educational, Scientific and Cultural Organization (UNESCO), the United Nations Children’s Fund (UNICEF), the UK Department for International Development (DFID), the United Nations Girl’s Education Initiative (UNGEI), the Civil Society Forum to End Violence against Children, the World Bank, Education Cannot Wait (ECW), the Global Partnership for Education (GPE), the Global Business Coalition for Education, and the Global Partnership to End Violence Against Children 41 World Bank (2020d). The five pillars are: (1) learners are engaged; (2) teachers facilitate learning; (3) learning resources are adequate and diverse; (4) schools are safe and inclusive; and (5) and systems are well managed. 7 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 various forms of VIAS (physical, sexual, and emotional) and A brief conclusion follows, including suggestions for further provides basic information on their prevalence. Violence research. There is a need not only to test innovative programs remains ubiquitous in schools. The issues of corporal to end VIAS in developing countries where the available punishment in school and the continuum of violence and risk evidence on what works remains limited, but also to conduct of poly-victimization are also briefly discussed. cost-benefit analysis of those interventions. We need to better incorporate violence-related indicators in rigorous The next two chapters provide tentative estimates of the evaluations assessing the impact of interventions on learning potential impacts of VIAS on schooling, learning, and outcomes. Another priority should be to improve the design earnings (Chapter 2), and on health and well-being, fertility, of surveys aiming to measure VIAS, so that apart from and agency (Chapter 3). The analysis relies on a wide range of measurement, the surveys can be used more systematically surveys. The analysis demonstrates that violence has lasting to provide insights on the correlates of VIAS and its impacts. negative impacts on students, including lower academic In addition, questions on VIAS could be integrated in other performance and poor physical, mental, and emotional surveys, such as multi-purpose surveys used for poverty health. It also undermines the ability of communities and analysis – these surveys often contain detailed modules on governments to create schools that are safe harbors in the education of household members, but rarely questions which children can develop their abilities and skills, while on VIAS. Adding such questions would not lead to much also embracing the values of cooperation, mutual respect, longer surveys, and could provide a richer data source for peaceful problem solving, inclusion, and gender equality. analysis than those currently available. Based on measures of potential impacts, estimates of potential lifetime economic costs especially for lost earnings Another area where research is needed is how to implement are provided (Chapter 4). reforms and scale up successful programs. Furthermore, the question of what to do to prevent VIAS is more complex The second part of the study is devoted to a discussion of than simply identifying interventions that can achieve programs and policies to end VIAS. A framework is first positive impacts. Efforts to end VIAS can be successful only provided on some of the main risk factors leading to VIAS with an understanding of the political economy of broader (Chapter 5). No single factor can predict violent incidents. reforms, as well as the enforcement of existing laws and Rather, it is often a combination of factors at the individual, regulations. Research is needed to better assess the social relationship, school, community, and societal levels that constraints to violence prevention, why some societies or leads to violence. The accumulation of risk factors often communities tolerate or even condone the use of violence explains why an individual behaves more violently or is more in various contexts, what incentives and accountability prone to be victimized than another. Therefore, instead mechanisms may be needed at different levels to end of looking for a single intervention to prevent violence, VIAS, and how policymakers should prioritize and sequence it is best to combine interventions to tackle salient risk programs and policies to enable all stakeholders to play an factors and enhance relevant protective factors. Promising active role in ending VIAS. This study does not cover all of school-based interventions are discussed within a life cycle these questions. But it does hopefully provide a few pointers approach (Chapter 6). In addition, to prevent VIAS, multiple on the potential benefits of ending VIAS, and promising stakeholders must be engaged within the school –students, approaches to do so. teachers, and principals, but also in in the community, including parents and caregivers (Chapter 7). Examples of effective programs are provided. Finally, illustrative cost- benefit analyses are reported showing that investing in interventions to prevent VIAS is likely to generate benefits well above the interventions’ cost (Chapter 8). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 8 PART I locations, not only (for children) at school, but also at home and in communities, including online communities. VIOLENCE IN SCHOOL Figure 1.1 provides a diagram with key concepts often used AND ITS IMPACTS when talking about violence against children. Child abuse is a broad category that reflects among others violence and neglect occurring to children under 18 years of age. It includes all types of physical and/or emotional ill-treatment, CHAPTER 1 sexual abuse, neglect, negligence and commercial or other exploitation, which results in actual or potential harm to TYPES OF VIOLENCE AND the child’s health, survival, development or dignity in the context of a relationship of responsibility, trust or power. PREVALENCE Exposure to intimate partner violence is also sometimes included as a form of child abuse. Abuse is a broader concept Violence is a ubiquitous problem in schools in many than violence per se, but all three main forms of violence – communities throughout the world, directly affecting sexual and emotional or psychological, and physical violence teachers and students and indirectly having an impact on contribute to abuse: the wider community itself. It has multiple, lasting impacts on students. It also undermines the ability of communities • Sexual violence is defined as any sexual act, intimidation, and governments to create schools that are safe harbors attempt to obtain a sexual act, unwanted sexual where children and adolescents can develop their abilities comments or advances against another individual and skills while also embracing the values of cooperation, using coercion. In addition to the lasting physical and mutual respect, peaceful problem solving, inclusion, and psychological damage that sexual violence frequently gender equality. This chapter first defines the various forms exacts on victims, it also can have subsequent and lasting of VIAS before providing information on their prevalence negative effects on victims’ health in terms of sexually using a range of data sets. The issues of corporal punishment transmitted diseases (STDs), and other health issues. in schools, the continuum of violence, the risk of poly- victimization, and the transmission of violence across • Emotional or psychological violence may include generations are also briefly discussed. verbal and emotional abuse, such as isolating, rejecting, ignoring, insulting, spreading rumors, making up lies, TYPES OF VIOLENCE name-calling, ridiculing, humiliating and threatening. It can be inflicted in schools by student as, but also A commonly accepted definition of violence comes from by teachers through forms of punishment that the World Health Organization’s 2002 World Report humiliate, denigrate, scapegoat, threaten, scare, on Violence and Health42 which, following an earlier or ridicule students, frequently again with long- consultation43 , defined violence as “the intentional use of term consequences for the children’s health and physical force or power, threatened or actual, against a person or well-being. While emotional violence may be less group that results in or has a high likelihood of resulting in injury, visible than physical or sexual violence, it can also death, psychological harm, maldevelopment, or deprivation.” have significant and lasting impacts on students. Violence is the result of an abusive use of force or power against a person. The harm can be actual or threatened and • Physical violence is any form of physical aggression it may take different forms. It can lead to injury or death, with intent to hurt another person. It can manifest but also to trauma or other mental health symptoms. As itself from student to student, student to teacher, or such, the consequences of violence may not always be teacher to student. One manifestation of teacher to highly visible, but they are nevertheless damaging. Violence student violence is corporal punishment which remains is also often multidimensional, meaning that individuals are used in many schools as a form of discipline mechanism often subjected to multiple forms of violence and in multiple to correct misbehavior in the classroom, or punish 42 Krug et al. (2002). 43 WHO Global Consultation on Violence and Health (1996). 9 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Figure 1.1: Forms of Violence in and Around Schools CHILD ABUSE Emotional or Sexual violence Physical violence psychological violence Intimate partner violence Bullying Corporal punishment School-related Gender Based Violence Source: Authors. children for poor academic performance or other in which physical force is used and intended to cause reasons. Physical violence is the most common form of some degree of pain or discomfort, however light, violence in schools and is highly visible to all students. as well as cruel and degrading non-physical forms of punishment. Corporal punishment remains highly In addition to those three broad categories or forms of prevalent in schools as a way to discipline students. violence, given the specific focus of this study on violence in and around schools, Figure 1.1 also highlights particular types • Intimate partner violence or IPV refers to any of violence which can themselves sometimes be expressed in behavior within an intimate relationship that causes multiple ways through physical, emotional, or sexual violence. physical, psychological or sexual harm to those in the relationship. Dating violence between students • Bullying is defined as unwanted, unwelcomed, is a form of IPV and VIAS, while sexual harassment repeated aggressive behavior among students, and is on the way to or from school by individuals who are among the most common forms of violence in school not a partner is a form of VIAS, but not of IPV. settings. As for other types of violence, bullying can have serious and lasting physical, mental and Figure 1.1 also highlights gender-based violence, which refers emotional problems for victims. In an era of rapid to acts or threats of sexual, physical, or emotional violence technological growth, and in the context of a higher occurring in and around schools, perpetrated as a result of emphasis placed on remote learning brought about gender norms and stereotypes, and enforced by unequal by the COVID-19 pandemic, cyberbullying (bullying gender dynamics. When considering gender-based violence which takes place over digital devices) has become in schools, it is important to understand the link between a front-and-center issue that schools, educators, violence against women and violence against children. This parents, and communities must struggle with. Bullying links persist through the intergenerational transmission of and cyberbullying are highly prevalent in schools. violence and the gendered continuum of violence. Children who have witnessed or experienced violence at home at an • Corporal punishment is also a common form of violence early age are more likely to either perpetrate violence or against children worldwide. It includes any punishment be victimized when they grow up. Child maltreatment is JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 10 particularly damaging for the child’s development, but simply Behavior in School-Age Children survey (HBSC) which witnessing violence is also damaging. Children who live in a includes mostly European countries as well as Canada and household where the mother or other women are chronic a few countries from the MENA region; and (3) and the victims of violence face higher risk factors for future violence Programme for International Student Assessment (PISA) or victimization. which includes mostly middle and high income countries. The other two surveys are implemented in the population as Different forms of violence reinforce each other and a whole: the Violence against Children Survey (VACS) and tend to feed a self-reproducing cycle. This is referred the Demographic and Health Survey (DHS), both of which to as the continuum of violence from the home to the have been implemented in low and middle income countries. streets to the school, where victimized children tend to Details are available in background notes. Estimates are experience violence in multiple places, often in multiple averages across countries. forms. The root causes for this continuum of violence include patriarchal systems that endorse the use and The proportions of children involved in fights or victims of abuse of power over others, installing a hierarchy among bullying are broadly similar in both the GHSH and HBSC forms of violence that tends to acknowledge the severity surveys for which questions tend also to be broadly of some (for example, when men are mostly victims) while comparable. In the PISA survey, the questions that are understating the impacts of others (for example, when asked are different46 , but the estimates of VIAS are of women and children are affected). As such, systematically the same order of magnitude as in the GSHS and HBSC looking at violence dynamics and risk factors with a gender surveys. The share of children affected by physical violence lens is crucial in identifying sustainable solutions. is a bit lower in the VACS surveys. For other forms of emotional violence and for sexual violence (not available in PREVALENCE OF VIOLENCE IN SCHOOLS GSHS and HBSC surveys), proportions are lower in the VACS, but due to limits in the data as well as the possibility Reports and studies44 have been published on the prevalence of underreporting, actual values could be larger. In the of various forms of violence, including in and around schools. DHS, sexual violence (being forced to have sex) likely to For violence in schools, analysis typically relies on multi- happen in school is limited, but may be underreported, for country school health surveys which focus on physical example when the topic is considered taboo. Unfortunately, violence and bullying, although other types of surveys can data on sexual harassment which affects mostly girls are not be used for specific countries (on data sources, see Annex widely available. 2). In a recent report on the prevalence of violence in schools globally, UNESCO finds that a third of students Overall, the data indicate that the prevalence of VIAS is high are bullied by their peers at school at least once in the last in most countries. This said, there needs to be some caution month and a similar proportion are affected by physical on differences in prevalence between countries since the violence. Among the subset of countries with trends data sets are limited in geographical scope and age groups over time, half of the countries had a decline in violence, included, and the measures used for different surveys are not one fifth had an increase, and there was no change over always the same and therefore not strictly comparable, even time in prevalence in the other countries. This suggests if the surveys have many commonalities. that without more forceful programs and policies, limited progress is likely to be achieved over time45 . Table 1.1 provides a summary of estimates prepared for this report on the prevalence of VIAS from a range of surveys. Three of the surveys are implemented in secondary schools: (1) the Global School Health Survey (GSHS) which includes mostly low and middle income countries; (2) the Health 44 See among others UNICEF (2014), Office of the SRSG on Violence against Children (2016), Hillis et al. (2016), UNICEF (2017, 2019), Know Violence in Childhood (2017), UNESCO (2019), and World Health Organization et al. (2020). 45 UNESCO (2019). 46 In PISA, children are asked if during the past 12 months, they (1) were left me out of things on purpose by other students; (2) other students made fun of them; (3) they were threatened by other students; (4) other students took away or destroyed things that belonged to them; (5) they got hit or pushed around by other students; and (6) Other students spread nasty rumors about them. Among those, (1) and (2) are not strictly speaking measures of violence and therefore are not reported here, but they are included as controls in subsequent regression analysis. 11 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Table 1.1: Prevalence of Violence in Schools in Multi-country Surveys (%) GSHS HBSC PISA VACS DHS GSHS, HBSC, and PISA surveys Attacked in last 12 months 37.8 - - - - Involved in fight in last 12 months 27.6 31.2 - - - Injured in last 12 months 31.3 44.5 - - - Injured from fight 1.5 1.8 - - - Bullied in last 30 days 29.5 29.0 - - - Others left me out of things Students made fun of me - - 23.8 - - Threatened by other students - - 26.5 - - Others destroyed my things - - 23.4 - - Hit by other students - - 33.5 - - Nasty rumors about me - - - - VACS and DHS surveys - - 28.7 - Physical violence in schools - - 3.5(*) - Emotional violence in schools - - 2.8(*) 1.5(*) Sexual violence in schools - - 2.8(*) 1.5(*) Source: Authors’ estimates. Note: The prevalence of emotional and sexual violence in VACS and DHS surveys may be underestimated. While we do not attempt to provide estimate of how the simple text, may increase by 10 percentage points under a COVID-19 pandemic may affect violence in school once pessimistic scenario53 . For schooling, girls are especially likely students return to school, the pandemic is likely to have to be affected54 , leading to higher risks of child marriage55 exacerbated some of the factors that lead to violence against with major implications for the rest of their life56 . More children in general, and violence in schools in particular. generally, children from vulnerable groups, including not only Chapter 5 discusses some of the risk factors for being a girls and those in extreme poverty, but also children with perpetrator or the victim of violence in school. Several of disabilities, refugees and internally displaced persons, and those risk factors may have been exacerbated by the current other groups are especially at risk. crisis. Many individuals and households are under stress due among others to social isolation, losses in employment and CONTINUUM OF VIOLENCE, POLY- income, and illnesses or death from the pandemic. Initial VICTIMIZATION, AND VULNERABLE GROUPS predictions of the economic impacts of the crisis were dire47 for both developed48 and developing countries49. Over time The various forms of violence often do not occur in isolation. many projections were further revised downward, and while Instead, they tend to reinforce each other. In addition a recovery is underway, the consequences of the crisis have children are often victims of violence in separate locales, at been severe. Estimates suggest that the crisis may lead to an school but also at home and in the community. This feeds increase in the number of poor people of 150 million50 . Of into a self-reproducing cycle57 which can be referred to as those, about half are children. According to the World Food the continuum of violence from the home to the streets to Programme, the number of people suffering from acute the school, where victimized children tend to experience hunger may have doubled51 . Student learning suffers during violence in multiple places and often through multiple forms recessions52, and estimates suggest that learning poverty as suggested in Figure 1.2. Poly-victimization is another – the share of children not able to read and understand a term used to refer to the fact that children may be affected 47 International Monetary Fund (2020). 48 For Europe, see European Commission (2020). 49 For sub-Saharan Africa, see World Bank (2020a). 50 World Bank (2020b). 51 Food Security Information Network (2020). School lunch programs were also affected. These programs serve many children (World Food Programme, 2013, 2020). 52 Shores and Steinberg (2019). 53 Azevedo (2020). 54 See UNDP (2015), Onyango et al. (2019), and Bandiera et al. (2019). See also World Bank (2020c) for a review, as well as Asfaw (2018) on Ethiopia, Dureya et al. (2007) and Cerutti et al. (2019) on Brazil, and Lim (2000) on the Philippines. 55 Wodon et al. (2016, 2017); Kassa et al (2019). 56 Wodon et al. (2018). 57 Wilkins et al. (2019). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 12 by violence in multiple locales, as well as by multiple forms physical, mental and emotional harm. Poly-victimization has of violence – not only physical violence, but also sexual a cumulative negative effect that leads to complex trauma, and emotional violence. In the United States, the National which requires specific attention as it signals a child’s greater Survey of Children’s Exposure to Violence shows that for a vulnerability and distress. Policymakers and stakeholders child who experienced physical violence in the past year, the working in school need to take poly-victimization into risks of being victim of sexual abuse and suffering from child account to respond to the multiple layers of risks and distress maltreatment are five and four times higher, respectively. of children and to target the most vulnerable children58 . Children’s poly-victimization has multiplier effects on As an illustration of poly-victimization at work in schools, their wellbeing and capacity to learn as children exposed Table 1.2 provides estimates of the marginal impact of to multiple forms of violence are at higher risk of lasting one type of violence on the likelihood of being the victim Figure 1.2: Continuum of Violence and Poly-victimization Continuum of violence PHYSICAL PSYCHOLOGICAL Incl. Corporal punishment SEXUAL BULLYING/ CYBER BULLYING Aggression and anti-social behaviors HARASSMENT Polyvictimization Source: Adapted from UN Women and UNESCO (2016). of another type of violence. For example, in the GSHS of violence. This is not only the case in school, but also at surveys, the coefficient of 0.169 in the last column of the home and in the community, leading to more serious trauma first row suggests that having been bullied is associated with for the children affected. an increase in the likelihood of having been attached by 16.9 percentage points. All coefficients in the Table are positive One of the root causes for this continuum of violence is the and statistically significant, suggesting positive associations patriarchal system that endorses the use and abuse of power between the various forms of violence and thus pointing to over others, including by installing a hierarchy among forms the fact that children are often the victims of multiple forms of violence that tends to acknowledge the severity of some 58 Finkelhor et al. (2011). 13 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Table 1.2: Poly-victimization – Estimates of Marginal Effects between Multiple Types of Violence Attacked Fought Injured Bullied GSHS Data Attacked - 0.268 0.140 0.169 Fought 0.256 - 0.143 0.122 Injured 0.127 0.136 - 0.138 Bullied 0.171 0.129 0.155 - HBSC Data Fought - - 0.094 0.114 Injured - 0.115 - 0.075 Bullied - 0.107 0.058 - Source: Authors’ estimation based on GSHS and HBSC surveys. (when men are mostly victims) while possibly understating gender conforming students and suicidal behavior. the impacts of others (for forms of violence that women and The prevalence of various forms of VIAS also differs by children mostly experience). In school, violence may also gender and age. For example, the prevalence of physical start with actions among students that may seem innocuous violence and bullying tends to decrease among older to educators, but they can escalate and lead to extreme students, although the severity of particular episodes of consequences such as death or suicide. Understanding this VIAS may in some cases increase. continuum is critical to act early to prevent episodes of violence from escalating or prevent them from happening. To conclude, violence is often a manifestation of underlying tensions within the family, the community, or the school. Some groups of children may be especially at risk of VIAS, As will be discussed in subsequent chapters devoted to in some cases with compounding effects when multiple programs and policies to end VIAS, in schools teacher risk factors are combined. Children who have witnessed or training is needed to ensure that schools are child-friendly experienced violence at home, including at an early age, tend and inclusive – this matters for educational attainment and to have a higher probability to either perpetrate violence or learning in general, but also to prevent VIAS. In communities be victimized when they grow up59. Violence in childhood is as well and within the family, interventions can help to particularly damaging, with consequences that can last for a reduce VIAS. When seeking solutions, this suggests the life time60 . Witnessing violence is also damaging as children. need to systematically look at violence dynamics and risk When the mother or another woman in the household factors, including with a gender lens, as will be discussed in is a victim of violence, children are at higher risk of both subsequent chapters63 . perpetrating violence or being a victim later in life61 . Some groups of children are especially vulnerable to violence, for example when they are mocked by other children or not seen as equals. This is the case for children with disabilities as well as various minorities – including (depending on the country) ethnic, racial, religious, LGBTQ, and indigenous peoples62. Some studies suggest a link between violence against non- 59 On LGBTQ, see for example Hotaling and Sugarman (1986); Bandura (1986); Fargo (2009). 60 Zolotor et al. (1999); MacMillan et al. (2009). 61 Bair-Merritt et al. (2006); Fulu et al. (2017); Jewkes et al. (2002); Meltzer et al. (2009); Guedes et al. (2016); Fry et al. (2018). 62 UNESCO (2017). 63 Jewkes et al. (2002); Heise (2011); Namy et al. (2017). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 14 CHAPTER 2 the student’s growth, resilience, as well as her openness and tolerance vis à vis others’ cultures and beliefs.65 By IMPACTS ON EDUCATION contrast, in contexts marked by violence, both in school and in the community or country, experiences in schools AND EARNINGS may contribute to distrust, so that specific support mental health or psychosocial support may be required to help students66 . In this section, the aim is to suggest estimates VIAS can have a wide range of negative impacts. This of the potential negative impact of violence in school on chapter provides evidence of potential impacts related to student’s experience in school. education64 (schooling and learning) and work (labor force participation, earnings, and productivity). For education , the GLOBAL ESTIMATES focus in this chapter is on (i) the student’s perception of his/ her experience in school or connectedness with the school Data are available on student’s perception of their experience and whether he/she missed days at school; (ii) the student’s in school in the HBSC, GSHS, and PISA surveys – the learning performance as measured through international main surveys to measure the prevalence of physical violence assessment data or grades in school; (iii) whether VIAS may and bullying in schools. It is, therefore, feasible to assess the be perpetrated by teachers or school principals for providing direct potential impact of VIAS on those perceptions. In education services, using the prevalence of petty corruption the PISA data which covers many middle and high income as a proxy; (iv) whether VIAS is a major reason for students’ countries, data are available on whether (1) students have no decision drop out of school; and (v) the relationship between close friends, (2) feel lonely, (3) feel that other students do VIAS and educational attainment. not accept them like they are, and (4) find it easy to make friends. Information is also available on whether students feel For potential impacts related to work in adulthood, the that their teachers (5) make them confident, (6) listen to focus is on (i) expected (wage) earnings when working; (ii) them, and (7) understand them labor force participation and the type of employment held; and (iii) perceptions of the household standard of living. With the HBSC surveys for European countries plus Canada It is important to emphasize that violence in schools can and a few countries in the Middle East and North Africa have both direct and indirect impacts on children. When region, information is available on whether students (1) like considering schooling and learning, direct potential impacts their school; (2) feel that their teachers think they are doing of VIAS are estimated. By contrast, when considering work- poorly academically in comparison to other students; (3) feel related outcomes, indirect potential impacts are estimated that other students do not accept them the way they are; (4) through educational attainment and learning. feel that their teachers do not accept them the way they are; (5) feel that their teachers do not care about them as a person; STUDENTS’ EXPERIENCE IN SCHOOL and finally (6) have no trust in their teachers. In most cases, multiple answers can be provided by students, for example, Violence in school affect the relationships between ranging from strongly agree to strongly disagree when asked students, and between students and their teachers. It can about various perceptions, but all outcome variables have also affect the community, or vice-versa, and reinforce been dichotomized (yes/no coding) for the analysis in order exclusion. This in turn can affect how students view their to facilitate the interpretation of the results. The data from own education, and how they perceive not only their the GSHS are less detailed, but three perceptions of students schools, but their teachers as well as whether they even related to their experience in school are available: (1) whether want to go to school. Simply said, violence in school affects the students have close friends in school; (2) whether they student’s socio-emotional well-being as well as their missed school days; and (3) whether they can benefit from socio-emotional skills (SEL), which in turn are critical for help from other students or not. 64 Students who are victims of VIAS are more likely to miss school, feel anxious, and lack friends and trust in teachers and principals. They may not be learning as well. Teachers working in violent environments are also often underpaid, overworked, and possibly fearful of being victims of VIAS themselves. They may lack the skills to manage violence. This may affect their teaching performance, and student learning. Because schools are embedded in communities, IAS may lead to violence and crime in communities, and vice versa. Failure to prevent VIAS can have lasting negative impacts. This chapter provides evidence of some of these potential impacts, focusing on students who are victimized as opposed to students who are perpetrators of violence or may have witnessed violence but were not the target of it. 65 As an example of analysis of student well-being with PISA data, see OECD (2015). 66 See Inter-Agency Standing Committee (2007). 15 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 To assess the direct potential impacts of VIAS on those missing school of 3.4 percentage points in the GSHS data. perceptions controlling for the factors that may affect For being involved in a fight in those data, having an injury, them, a range of controls are included in the regression and being bulled, the negative effects at the margin are larger analysis67. Table 2.1 provides the results for the potential at respectively 8.9, 7.8, and 5.8 percentage points. Note impacts of the variables of interest (VIAS) with the various that for injuries, we consider all types of injuries, whether surveys. Note that the HBSC data includes information on likely to have taken place in schools due to physical violence whether students have been involved in a fight, whether they or not, but only a minority of injuries are due to violence have been injured (not only because of violence, but also even if all injuries may affect the outcomes of interest. because of other reasons such as being engaged in sports When conducting simulations of the potential impacts of or work), and whether they have been bullied. In addition, violence based on the regression results, only injuries likely the GSHS data has information on whether students have to be directly related to physical violence in school are been the victim of a physical attack. In the PISA data, we accounted for. focus on whether children were threatened, had their things destroyed, were hit, or were the subject of nasty rumors. Most coefficients in Table 2.1 are statistically significant regardless of the dataset, suggesting systematic negative The interpretation of the coefficients is in terms of potential impacts of VIAS. There are a few exceptions. percentage points at the margin. For example, controlling For example, in the GSHS dataset, physical fights are for other factors, being the victim of a physical attack is associated with a slightly lower likelihood of having no friends associated at the margin with an increase in the probability of (perhaps because some fights were to support friends), Table 2.1: Marginal Potential Impacts of VIAS on Students’ Experience in School, GSHS and HBSC GSHS Data Physical Attack Physical Fight Injured Bullied No close friends in school 0.008 -0.004 NS 0.011 Missed school 0.034 0.089 0.078 0.058 Not helped by other students 0.023 0.015 0.01 0.051 HBSC Data Physical Attack Physical Fight Injured Bullied I do not like my school at present - 0.083 0.016 0.021 My teachers think I am not doing well academically - 0.081 0.022 0.027 Other students do not accept me like I am - 0.014 NS 0.165 My teachers do not accept me like I am - 0.073 0.021 0.038 PISA Data Threatened Things Destroyed Hit by Others Nasty Rumors I have no close friends NS NS NS 0.004 I feel lonely 0.031 0.007 0.006 0.041 Other students do not accept me like I am 0.03 NS NS 0.048 It is not easy to me to make friends 0.013 NS NS 0.012 My teachers make me confident 0.012 0.018 0.01 0.032 I feel my teachers listen to me 0.012 0.018 NS 0.033 I feel my teachers understand me 0.013 0.019 NS 0.037 Source: Authors’ estimations using GSHS, HBSC, and PISA data. NS means statistically not significant at 0.1 level. 67 For the HBSC surveys, the controls include gender, age, grade, the student’s height, whether the student is underweight or overweight, whether the student participates in physical education and for how many days per week, whether the students feels that other students are kind and helpful, whether the household in which the student lives is well off as per the student’s assessment (with various degrees of being well off), whether the mother and father of the student live at home, whether the student finds it easy to talk to his father and mother, whether the student feels that friends and family try to help when in need. Some of these variables could perhaps be considered as outcomes as well, but they are used for the analysis as controls. For the GSHS data, the controls include gender, age, grade, the student’s height, whether s/he goes hungry, whether s/he is underweight or overweight, whether s/he is active, whether s/he benefits from physical education at school, the time spent sitting without activity, whether parents check his/her homework, understand his/her trouble, know what s/he does in his/her free time, and go through the child’s things. Other factors could affect the outcomes of interest, but these are the variables available in the dataset that seem to be the most important potential factors affecting outcomes. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 16 and the association between injuries and perceptions that simple average across countries without country population other students do not accept the student as they are is not weights, and next an average accounting for the difference statistically significant, suggesting no direct impact in that between countries in the size of their student body. case (indicated by NS in the Table). Still, by and large, the Considering again missing days of schooling as an example, effects are systematic. the share of the students missing school is 28.4 percent without country weights and 20.0 with country weights. If all What might be the aggregate potential impact on these forms of VIAS were eliminated, that share would drop by 5.6 outcomes if VIAS were eliminated? The potential benefits percentage points without country weights and 4.0 points from ending VIAS depend on both the sign and magnitude with country weights. In other words, the simulations suggest of the marginal impacts estimated through the regression that one in six instances of missing school may be due to analysis, and the share of students experiencing each form of VIAS. For other outcomes, the proportions are lower, but VIAS. In Table 2.2, the first two columns provide the baseline still substantial. This is the case with the other surveys as well. predicted values of the various outcomes, considering first a Table 2.2: Simulations of Potential Impacts of Ending VIAS on Student’s Experience, GSHS and HBSC Predicted Share Under Potential Impact of VIAS Share Associated with VIAS Baseline Conditions Simple Weighted Simple Weighted Simple Weighted Average Average Average Average Average Average GSHS Data No close friends in school 6.1 6.6 0.5 0.4 7.2 5.2 Missed school days 28.4 20.0 5.6 4.0 16.7 14.9 Not helped by other students 25.0 22.6 2.6 2.0 10.3 8.3 HBSC Data I do not like my school at present 24.4 25.6 3.1 3.2 10.9 10.8 My teachers think I am not doing well 32.4 33.7 3.1 3.2 8.7 8.7 Other students do not accept me like I am 23.4 25.1 4.8 5.0 14.7 14.9 My teachers do not accept me like I am 19.5 20.6 3.3 3.5 14.3 14.3 My teachers do not care about me 36.6 38.4 2.1 2.1 5.6 5.4 I do not feel a lot of trust in my teachers 38.5 40.3 2.8 2.8 7.4 7.2 GSHS Data I have no close friends 1.6 1.7 0.2 0.3 9.2 9.8 I feel lonely 18.1 18.9 2.8 3.0 10.1 10.7 Other students do not accept me like I am 24.4 21.6 2.0 2.2 6.8 7.6 It is not easy to me to make friends 23.0 23.2 0.7 0.7 2.5 2.7 My teachers make me confident 22.2 21.5 1.7 1.7 7.5 7.9 I feel my teachers listen to me 26.7 26.8 1.6 1.7 6.1 6.3 I feel my teachers understand me 27.3 26.4 1.8 1.8 6.1 6.5 Source: Authors’ estimations using GSHS, HBSC, and PISA data. Note: The simple average is an average values across all countries without considering differences in student population between countries. The weighted average factors in the countries’ student population. 17 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 UNITED STATES of the coefficients is again in percentage points. For example, the estimate for fights and trust in the school’s processes for Analysis is also conducted using the School Survey on Crime the 2015 survey suggests that students involved in fights are and Safety (SSOCS) for the United States, a rich dataset in 10.3 percentage points more likely to not have thrust in the terms of the types of violence included since information is school’s administrative processes including those to handle available on seven different types of VIAS or proxies thereof: violence. The estimate obtained with the 2015 survey is fights, bullying, hate, avoidance strategies, fear, weapons, and similar. In virtually all cases, exposure to VIAS leads to worse finally gangs. The first six variables are related to victimization outcomes. The only exception is whether VIAS affects the and behaviors by individual students while the last variable likelihood of pursuing one’s education beyond high school (in relates to the presence of gangs in the school, whether most cases impacts that are not statistically significant). individual students are involved in them or not. In terms of outcomes, five variables are considered, namely whether How much difference might ending VIAS make? Table 2.4 students (i) trust the school’s administrative processes provides the simulation. The potential impacts tend again (including to deal with VIAS); (ii) feel adults and teachers do to be large. For example, with the 2015 data, the share not care about them; (iii) lack friends; (iv) expect to continue of students who do not trust the school’s administrative their education beyond high school; and (v) know how to processes could be reduced by 7.0 percentage points or get drugs in the school. The analysis controls for selected about a fourth of the base value if VIAS were eliminated. other factors that may affect these outcomes, including The magnitude of the proportional reduction from the base gender, age, race, the grade the student is in, the number of is similar for the share of students who feel that adults and school activities that the student is involved in, the number teachers in the school do not care about them, feel they lack of security measures in place in the school, and a measure of friends, or know how to get drugs in the school. The impact household income (series of income levels). by contrast on the share of students who are considering pursuing their education beyond high school is virtually non- Table 2.3 provides estimates of the marginal potential existent. The fact that results are similar with the 2015 and impacts of VIAS with two years of data. The interpretation 2017 surveys is indicative that effects may be systematic. Table 2.3: Marginal Potential Impacts of VIAS on Student’s Experience in the Classroom, United States Fights Bullied Hate Avoidance Fear Weapon Gangs 2015 Survey No trust in school 0.103 0.157 0.152 0.107 0.085 0.164 0.145 Teachers not caring 0.047 0.032 0.044 NS 0.035 0.032 0.023 Lack of friends 0.051 0.024 NS 0.047 NS NS 0.035 Studying beyond high school NS NS 0.041 NS NS NS NS Knows how to get drugs 0.138 0.233 0.221 0.099 0.117 0.371 0.358 2017 Survey No trust in school 0.105 0.185 0.115 0.085 0.057 0.095 0.226 Teachers not caring 0.062 0.051 NS NS 0.039 NS 0.037 Lack of friends 0.032 0.033 0.034 0.029 0.047 NS NS Beyond high school NS NS NS NS NS NS -0.027 Knows how to get drugs NS 0.192 0.193 0.089 0.121 0.276 0.335 Source: Authors’ estimation using SSOCS data for 2015 and 2017. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 18 Table 2.4: Simulations of Potential Impacts of ending VIAS on Students’ Experience, United States Baseline Absolute Change Percentage Change (Predicted) with no VIAS with no VIAS (%) 2015 Survey No trust in school 26.7 -7.0 -26.4 Teachers not caring 6.7 -1.8 -26.2 Lack of friends 6.3 -1.4 -22.8 Beyond high school 85.6 -0.3 -0.3 Knows how to get drugs 41.3 -10.5 -25.4 2017 Survey No trust in school 26.4 -8.0 -30.3 Teachers not caring 7.4 -2.1 -28.1 Lack of friends 6.9 -1.9 -27.0 Beyond high school 86.0 1.1 1.3 Knows how to get drugs 38.4 -9.3 -24.3 Source: Authors’ estimation using SSOCS data for 2015 and 2017. LEARNING IN SCHOOL of teacher interactions with students can be beneficial for students well-being as well as for their learning72, but violence Violence in school may also affect student learning, in school may prevent such quality interactions. Following especially for vulnerable groups including girls. In some the analysis of the previous section, the aim of this section cases, teachers may be the perpetrators of violence instead is to estimate the potential negative impact of violence in of supporting for safe learning environments. Qualitative school on student learning. studies have pointed to relationships between VIAS and educational outcomes. In the Northern Triangle of Central PISA America, gangs generate climates of generalized fear that affects the quality of relationships and student’s academic Data from the PISA Assessment can be used to assess performance68 . A qualitative study for Bangladesh found the potential impact of violence in school on students’ that teachers and other adults suggests that experiences test scores. The assessment is implemented among 15 of corporal punishment could be a factor leading to school year-old children to assess their ability to use their reading, dropout and thereby a loss in learning69. In the UK, life mathematics and science knowledge and skills to meet real- histories suggest that a negative relationship between abuse life challenges. The data sets include a wide range of variables learning outcomes70 . Violence in school is also associated that can be used as controls when estimating the association with various types of disruptions in the classroom. Some between violence in school and student performance. As students may be deeply affected, while others may simply mentioned in Chapter 1, children were asked if during the be distracted, but in both cases this can have negative past 12 months, they (1) were left out of things on purpose implications for their ability to concentrate. Teachers by other students; (2) other students made fun of them; (3) may need to spend more time managing disruptions in they were threatened by other students; (4) other students the classroom, or may resort to classroom management took away or destroyed things that belonged to them; (5) practices correlated with lower academic performance they got hit or pushed around by other students; and (6) for students. The level of disruption in classrooms can other students spread nasty rumors about them. Among be measured, and studies suggest that such disruptions those variables, items (1) and (2) are not strictly speaking are associated with lower student performance, in turn measures of violence, but they are included as controls. suggesting that improving the classroom disciplinary climate may boost student performance71 More generally, the quality 68 Ranieri (2019). 69 Mitu et al. (2019). 70 Nelson (2009). 71 Cahu and Quota (2019). 72 See Hallinan (2008) and Baafi and Kwabena (2020). 19 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Key results for the potential impacts violence of are provided with a reduction in performance on the mathematics in Table 2.5. Test scores are scaled so that the international assessment of -22.6 points (-21.7 points for boys and -24.0 average for the ten countries is 500 points and the standard points for girls). Effects for the reading and science tests are deviation is 100 points when weighting all countries equally. similar. Other types of violence are also negatively associated This implies that two in three pupils are in a range of 400 with student performance, although the magnitude of the points to 600 points, and most students are in the 250 to effects is smaller. The negative effects of the proxies for 750 range. An average score of 500 does not mean that a VIAS are typically as large, and in some cases larger than the student is doing well, as many students do not achieve basic potential impact of a wide range of other variables on student language and math proficiency. As shown in Table 2.5, when performance, suggesting that VIAS matters. a student is threatened by other students, this is associated Table 2.5: Potential Impact of VIAS on Student Learning, PISA Mathematics Reading Sciences All Boys Girls All Boys Girls All Boys Girls I was threatened by other students -22.6 -21.7 -24.0 -28.2 -26.8 -30.2 -24.3 -23.0 -26.1 Others destroyed my things -0.9 -3.5 1.6 -5.3 -7.3 -2.9 -2.3 -3.9 -0.6 I got hit by other students -9.9 -10.5 -9.6 -14.8 -15.6 -14.4 -10.9 -11.4 -10.7 Others spread nasty rumors about me -3.2 -3.2 -2.1 -3.1 -4.4 -1.5 -3.4 -4.8 -1.4 Source: Authors’ estimations using PISA data. Would ending VIAS make a large difference in PISA scores mathematics and science are of a similar order of magnitude, at the national level? Table 2.6 provides the results of albeit a bit smaller, with gains of 2.1 percent from the base for simulations in which the four types of violence (listed above mathematics and 2.3 percent for science. In some countries, and captured in the PISA) are eliminated. Average reading gains are larger, while in other countries gains are lower. These scores across countries would increase by 11.9 points for gains may look small, but improving learning is not easy, and boys and girls combined, which represents an increase in gains in simulations for the potential impacts of other types performance of 3.1 percent from the base. Estimates for of variables or interventions are often not larger. Table 2.6: Simulations of Potential Impacts of Ending VIAS on Learning Performance, PISA Gain (absolute value) Gain (proportion from base, %) Boys Girls All Boys Girls All Reading (70 countries) 12.8 11.2 11.9 3.3 2.9 3.1 Mathematics (70 countries) 9.1 7.8 8.5 2.3 2.0 2.1 Sciences (70 countries) 10.2 8.9 9.5 2.5 2.2 2.3 Source: Authors’ estimations using PISA data. PASEC IN FRANCOPHONE AFRICA violence including corporal punishment by teachers. Almost two-thirds of students reported being beaten by teachers Proxies for VIAS are also available in the PASEC data for and one-third reported that other children don’t play with primary schools in ten Francophone African countries. The them, or that they are scared in school. Finally, teachers latest available data are for 2014 (only parts of the data for are asked whether they experienced physical, emotional, or the 2019 assessment are currently available). The PASEC sexual harassment, and how they perceive the school climate dataset also includes information on corporal punishment and security in the schools. by teachers –whether teachers report using it, and whether children state that they are beaten by teachers when they Key results are provided in Table 2.7. As for PISA, scores are punished or are perceived by teachers as not performing are scaled so that the average for the ten countries is 500 well. Students are asked whether other children play with points. The analysis is for students in grade 6. When a them (a negative response may be a good predictor of student reports not playing with others, this is associated bullying, at least for some students) and whether they feel with a reduction in score of up to 17.3 points depending on scared in the classroom/school, which could be a reflection of the student’s gender and the type of test. Students who JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 20 report feeling scared also score lower may suffer from a loss economic background of the student, the effect of either a of up to 23.0 points at the margin72. How do these effects hearing or visual disability, and many other factors affecting compare with the potential impact of other variables? learning such as teacher absenteeism, the level of education of teachers, or some of the characteristics of the schools. The negative effects of the proxies for VIAS are typically This suggests that VIAS may have large negative effects larger than the potential impact of variables on the socio- on learning. Table 2.7: Potential Impact of VIAS on Student Learning, 10 countries in Francophone Africa Mathematics Reading All Boys Girls All Boys Girls Not playing with others - Totally agree -16.5 -17.3 -15.5 -12.1 -12.1 -12.1 Not playing with others - Agree -10.3 -9.9 -9.3 -9.1 -8.7 -9.3 Feeling scared - Totally agree -21.6 -23.0 -19.2 -20.4 -21.8 -18.7 Feeling scared - Agree -17.0 -16.6 -16.6 -18.5 -19.3 -17.7 Corporal punishment by teachers 5.7 7.2 4.2 5.0 7.0 3.4 Source: Authors’ estimations using PISA data. Would ending VIAS make a large difference in PASEC are of a similar order of magnitude, with gains of 6.4 points scores at the national level? Table 2.8 provides the results of overall (5.1 points for boys and 7.5 for girls), which represents simulations in which violence and corporal punishment by an increase in performance of 1.24 percent from the base (1.1 teachers have been eliminated. Average reading scores for all percent for boys and 1.6 percent for girls). As was found for student in the 10 countries would increase by 5.7 points (5.2 the PISA assessment, in some countries gains are larger, while points for boys and 6.4 for girls), which represents an increase in others they are lower. But while impacts in PASEC are in performance of 1.2 percent from the base (1.1 percent for slightly smaller, their order of magnitude is similar to what was boys and 1.3 percent for girls). Estimates for mathematics observed in the analysis of PISA. Table 2.8: Simulations of Potential Impacts of Ending VIAS on Learning Performance, PASEC Gain (absolute value) Gain (proportion from base, %) Boys Girls All Boys Girls All Reading (10 countries) 5.2 6.4 5.7 1.1 1.3 1.2 Mathematics (10 countries) 5.1 7.5 6.4 1.1 1.6 1.4 Source: Authors’ estimations using PASEC data. Another interesting feature of PASEC is that teachers are One of the strategies for ending VIAS consists in providing asked whether they have experienced harassment, which appropriate pre-service and in-service training to teachers is a form of violence, and their perceptions of the school on how to ensure that schools remain safe. Another result climate in general, and specifically in terms of security at from the PASEC analysis worth mentioning is the fact schools. Some teachers state that they have been harassed that unfortunately, such training is rarely provided in emotionally, but the prevalence of physical and especially Francophone Africa, and probably in low and lower-middle sexual harassment is lower. A small share of teachers also income countries more generally. The data suggest that mention a lack of security at school and a negative school training on child-friendly and inclusive schools, the topics climate. Harassment of teachers in particular affects teacher most closely related to preventing violence in school, are satisfaction with their working conditions, which in turn the two categories of in-service training provided the least may affect the school climate and how well children learn in to teachers. It is also worth noting that Afrobarometer data school as measured by student assessments. for three dozen African countries suggest that one in five individuals are affected by petty corruption, which may lead to bribes, gifts, or favors, but may also in some cases to sexual violence against children. 72 Students who declare being victims of corporal punishment do actually slightly better, but effects are smaller (the largest effect is at 7.2 points). This positive effect is no reason to endorse corporal punishment given the possibility of other negative effects and the fact that the practice may contribute to perpetuating a culture of violence in schools and more generally in communities. 21 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 UNITED STATES Being involved in fights also increases the likelihood of being often distracted when doing schoolwork, and decreases For the United States, analysis is conducted using the the likelihood of being never or rarely distracted. Again, in SSOCS data mentioned earlier. In terms of learning virtually all cases, exposure to VIAS leads to worse outcomes outcomes, two variables are available: students’ grades for both grades and being distracted doing schoolwork. with A being the top grade and F implying failure, and whether students are distracted in class from doing their As before, the question of how much difference ending schoolwork because other students are misbehaving. For VIAS might make is explored through simulations whereby both outcomes, multiple categories are available, so ordered VIAS is eliminated and the predicted potential impact is logit models are used for estimations. The analysis controls measured. Table 2.10 provides the results. Ending VIAS for gender, age, race, the grade the student is in, household could increase the share of students getting an A by 2.4 income, and other factors that could affect these outcomes. percentage points, with a corresponding reduction in the Table 2.9 provides the key results. The interpretation of share of students receiving lower grades. For distractions the coefficients is slightly more technical, but a positive during schoolwork, the effects are a bit larger. Ending VIAS (negative) coefficient implies an increase (decrease) in the could increase the share of students never or almost never likelihood of observing that outcome. For example, being distracted by 7.9 percentage points, with a corresponding involved in fights decreases the likelihood of receiving an reduction in the other shares. A, and increases the likelihood of receiving lower grades. Table 2.9: Marginal Potential Impacts of VIAS on Grades and Distractions in Schoolwork, United States Fights Bullied Hate Avoidance Fear Weapon Gangs 2015 Survey Exam grades F's 0.001 0.001 NS 0.001 0.001 NS 0.001 D's 0.004 0.003 NS 0.003 0.003 NS 0.003 C's 0.036 0.030 NS 0.032 0.031 NS 0.0313 B's 0.049 0.041 NS 0.043 0.043 NS 0.042 A's -0.090 -0.075 NS -0.079 -0.078 NS -0.078 Distracted Never -0.081 -0.132 -0.090 -0.050 -0.107 -0.075 -0.086 Almost never -0.065 -0.105 -0.071 -0.040 -0.085 -0.060 -0.068 Sometimes 0.113 0.184 0.125 0.070 0.150 0.104 0.119 Most of the time 0.033 0.053 0.036 0.020 0.043 0.030 0.034 2017 Survey Exam grades F's 0.002 0.001 NS 0.001 0.001 0.001 0.001 D's 0.007 0.003 NS 0.003 0.003 0.003 0.002 C's 0.066 0.033 NS 0.030 0.034 0.027 0.020 B's 0.120 0.059 NS 0.054 0.061 0.049 0.036 A's -0.194 -0.096 NS -0.088 -0.099 -0.080 -0.059 Distracted Never -0.049 -0.162 -0.070 -0.065 -0.082 -0.075 -0.099 Almost never -0.034 -0.114 -0.049 -0.046 -0.057 -0.052 -0.069 Sometimes 0.066 0.217 0.093 0.087 0.109 0.100 0.132 Most of the time 0.018 0.059 0.026 0.024 0.030 0.027 0.036 Source: Authors’ estimation using SSOCS data for 2015 and 2017. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 22 Table 2.10: Simulations of Potential Impacts of Ending VIAS on Grades and Distractions, United States Absolute Change with Percentage Change Baseline (Predicted) no VIAS with no VIAS (%) 2015 Survey Exam grades F's 0.5 -0.1 -15.7 D's 1.5 -0.2 -14.9 C's 13.8 -1.5 -10.5 B's 42.4 -0.7 -1.6 A's 41.8 2.4 5.9 Distracted Never 18.7 3.9 21.0 Almost never 34.0 4.0 11.7 Sometimes 39.7 -4.8 -12.1 Most of the time 7.5 -3.1 -41.5 2017 Survey Exam grades F's 0.4 -0.1 -25.8 D's 1.3 -0.3 -24.4 C's 12.2 -2.1 -17.4 B's 41.6 -1.5 -3.6 A's 44.5 4.0 9.1 Distracted Never 20.7 4.8 23.2 Almost never 35.1 4.6 13.0 Sometimes 36.6 -5.7 -15.4 Most of the time 7.6 -3.7 -48.6 Source: Authors’ estimation using SSOCS data for 2015 and 2017. 23 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 REASONS FOR DROPPING OUT OF SCHOOL domestic or agricultural work; (vi) Have to do paid work to earn money (including agricultural work VIAS may also lead some students to drop out of school. other than at home); (vii) Need to learn a trade/skill Measuring the potential impact of VIAS on educational so working (including apprenticeships); (viii) Need to assessment is difficult as school surveys are not adequate stay home to look after siblings; (ix) Family member for this purpose since information is available only for ill, disabled, or elderly (providing care); and (x) Want students still in school. However, using other surveys, various to start working. In addition, the following modalities approaches can be used to assess whether VIAS leads related to transportation are also included in the children to drop out of school and how large the potential affordability category since lack of transport/distance impacts may be. One such approach relies on reasons stated often reflect a lack of resources in the household (such by parents as to why their children dropped out of school or as funds for boarding) to find alternative ways to go is not in school. In some countries, the survey questionnaires to school. Three reasons are related to distances to ask specifically if violence in school was the reason for school and transport: (i) Lack of transport; (ii) School dropping out. In particular, insights can be gained from the too far from home; and (iii) School not accessible Young Lives survey whose questionnaires include up to 30 for seasonal reasons (e.g., river prohibits access). modalities (depending on the country) as potential reasons for dropping out of school. Findings for Ethiopia, India, and • Child marriage and gender-based social norms: Vietnam for children aged 9-17 are provided in Table 2.11 These response modalities are especially salient for (for Peru, the same data in the survey are not available). The girls, but occasionally also affect boys. They include: data from the modalities have been summarized in a few (i) Marriage; (ii) Need to stay at home to look after aggregate categories for easier interpretation as follows: siblings; and (iii) [Parents considering it] not appropriate for girls to go to school or continue at school. • Violence and exclusion in school and elsewhere: Several of the response modalities are related to VIAS • Low quality of schools and poor academic performance or proxies for VIAS and other forms of violence. The of students: The modalities include: (i) Poor quality of modalities include (i) Bullying/abuse from peers; (ii) education at school (teaching and learning); (ii) Poor Banned from school for behavior reasons (which could quality of care (food, non-educational care); (iii) No include violent behaviors or fights); (iii) Ill-treatment/ need for schooling for future job (which suggest that abuse from teachers/principal; (iv) Not safe on the learnings is low or not useful); and (iv) Banned from way to school; (v) Child not welcome due to ethnic school due to failure to achieve necessary grade/level. group, caste, etc. (stigma and discrimination can be considered as a form of violence and therefore are • Other reasons: Other reasons included in the surveys included – the share of children affected is however include the following: (i) Disability or illness of the low); and (vi) Family issues and problems at home such child; (ii) Banned from school because away for too as parent disputes and marital conflict (this is included long (this could reflect an illness, injury, or other issues because it relates to the risk of violence at home). – including an inability to pay school fees); (iii) Child too young; (iv) Child still in pre-school (despite the age • Truancy and lack of interest: Truancy is the action bracket used for the estimations); (v) Refusal to answer; of staying away from school without a good reason. (vi) Not known; and (vi) Question not applicable. In other words, it is unjustified absenteeism. In Table 2.11, for violence in schools and elsewhere, the first • Lack of affordability of schooling and lack of transport: modality for responses (truancy, child does not want to Many of the response modalities relate to affordability go, not interested) is separated from the others because it issues due to out-of-pocket and opportunity costs. includes lack of interest and thus is likely to (vastly) overstate These modalities include: (i) Fees too expensive; (ii) the potential impact of VIAS. The main reason for not being Transport too expensive; (iii) Books and/or other in school full-time is lack of affordability for boys as well as supplies too expensive; (iv) Shoes, clothes, or uniform for girls in India and Ethiopia. This factor is also prominent for school too expensive; (v) Needed at home for in many other surveys with fewer modalities for responses. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 24 Truancy or the fact that the child does not want to go to One factor that may lead to relatively small shares of school or is not interested is also a major factor for not being parents stating that violence in school is a reason for not in school full-time, especially for boys, and in Vietnam it is being in schools is the normalization of violence. Parents the dominant factor. Gender issues, namely marriage and may consider use of corporal punishment as a legitimate gender norms which are forms of structural violence on girls way to discipline children and may not fully assess the and can also involve direct physical, sexual and emotional consequences on the wellbeing and health of girls and boys. violence, do affect girls. Note that because the sample of Similarly, bullying can be interpreted as a ‘normal’ experience children included is from 9 to 17 years, with most girls who in the schooling life of a child. As such, and given qualitative marry as children doing so at 16 or 17, the estimates over the research showing the widespread acknowledgement that larger age group tend to not fully reflect the pressure that different forms of violence happen in and on the way to adolescent girls have to marry early especially in India and schools, such results may underestimate the potential impact Ethiopia despite progress in reducing the practice of child of violence ion school. Still, overall, the estimates from the marriage in both countries. The quality of the schools is less Young Lives Surveys suggest that VIAS play a role for drop of a factor. This does not mean that the schools are of good outs, but is not the main factor. This finding is based on quality, but simply that other factors play a more important parental responses on reasons why their children dropped out role. One could also argue that poor quality is probably of school, but it is corroborated by other data that include leading to lack of interest in pursuing their education for questions to children themselves or to teachers and principals many children. The “other” category is very large, although as to the reasons for drop-outs. As an illustrative example, for different reasons depending of the country, and it seems consider data for Egypt where questions are asked to youth to reflect data issues in some of the countries. as to why they dropped out or did not pursue their studies, and Uganda where questions on the reasons children/youth Factors explicitly related to VIAS account only for a small drop out are asked to teachers and principals: share of the reasons for not being in school full-time in the Young Lives Surveys. The estimates are at 1.2 percent, 5.3 • Questions asked to youth: In Egypt, questions are percent, and 1.9 percent for India, Ethiopia, and Vietnam. asked to youth in the 2014 Panel Survey of Young But other reasons for not being in school full-time, especially People in Egypt (SYPE) as to why they did not pursue the truancy/lack of interest category, may be due in part to their education. More than 20 potential reasons are VIAS. If one were to assume that about one-tenth of the provided in the questionnaire. Two of those reasons truancy/lack of interest reasons actually reflect issues related relate to VIAS, namely (i) Maltreatment from teachers to VIAS, then the proportions of drop-outs related to VIAS or corporal punishment/Disciplinary reasons; and (ii) would increase to 3.0 percent in India, and 6.2 percent for Bullying. The other reasons are (in no particular order): both Ethiopia and Vietnam. In Chapter 5, when considering I finished my studies; Lack of schools for further study; the economic costs of VIAS related to children dropping out School fees; Private tutoring fees; The school is too of school, we will assume that across countries, about five far; I had to help out at home; I had to work; Helping a percent of children may be dropping our due to VIAS. This family member in his/her work; My parents didn’t want is admittedly a rough estimate, but not an unreasonable one me to go to school; My father does not want me to go given the available data. to school; I didn’t want to finish school; Not doing well in Table 2.11: Reasons for Not Being Full-time in School, Young Lives Data (%) India Ethiopia Vietnam Girls Boys All Girls Boys All Girls Boys All Violence 0.0 2.5 1.2 5.5 5.2 5.3 2.4 1.3 1.9 Truancy and no interest 26.3 9.8 17.8 9.8 8.8 9.5 46.2 38.2 42.8 Affordability 45.6 44.2 44.9 19.6 20.9 20.4 9.4 16.7 12.4 Gender 4.4 0.6 2.5 7.1 6.5 6.9 5.1 3.0 4.2 Quality 0.6 11.0 5.8 0.3 1.6 0.9 0.3 3.4 1.6 Others 23.1 31.9 27.7 57.7 57.0 57.2 36.6 37.3 37.1 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Authors’ estimations using Young Lives data. 25 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 school; To avoid mixing with the opposite sex; Marriage; important role in the decisions by children to drop out of For health reasons; Habits and traditions; Passed the school or to not continue their education. But the result is school enrollment age; Other (specify); Refused to important for estimations of the potential cost of inaction answer. The main reasons invoked for dropping out or towards VIAS (as discussed in Chapter 5), and for assessing not pursuing one’s studies are the fact that students the potential benefits to costs ratios for investments in completed their studies (70.2 percent of responses), interventions to prevent VIAS (as discussed in Chapter 8). and the fact that they did not want to finish school (13.0 percent). Other reasons are less frequently Before concluding this section, one last caveat should mentioned. Maltreatment from teachers or corporal be noted not to underestimate the potential impact punishment/Disciplinary reasons is mentioned by 0.7 of VIAS as a factor leading to drop-outs indirectly. percent of youth, and bullying is mentioned by less Consider West Africa. In Niger, the 2015 ENISED (Étude than 0.1 percent. VIAS could however lead students nationale d’évaluation d’indicateurs socio-économiques et not to be interested in furthering their education. démographiques) survey includes 16 potential responses for dropping out of school or not continuing one’s education: Too • Questions asked to teachers/principals: In Uganda, old or too young; [Enough] schooling completed; No school the question asked in national surveys to parents on or too far away; Family did not want more schooling; Cost the reasons why their children dropped out typically too high; Preference to work; Agricultural work; Domestic do not include modalities related to VIAS. However, a work; Pregnancy; Marriage; Schooling not adapted; Not question asked to principals in the Community Facility useful/no interest; Illness or handicap; Failure at exam; No Questionnaire of the Uganda National Panel Survey employment prospect from schooling; and Other. None of implemented from 2009 to 2011 does implicitly, these modalities refers to VIAS. Yet this does not mean that since discipline is one of the factors and it could be VIAS plays no role in children dropping out. Child marriage considered as a proxy for the risk of perpetrating (and early pregnancies) are a key reason for girls to drop VIAS (although other behaviors are likely to be out and they may be related to the school environment75 . included as well). At the secondary level, according Indeed, qualitative work that suggests that some parents do to 6.9 percent of principals, discipline was the main not send their adolescent girls to school because they fear factor leading boys to drop out of school. For girls, they may become pregnant in school – this is the case in the proportion was 2.5 percent. This suggests that in Niger, but also in other countries, including Burkina Faso and Uganda, violence may be a factor for dropping out. Ghana76 . The risk of sexual harassment on the way to school For boys, lack of interest, cost, and searching for a job or back from school is also real in many countries, even if the were the top three reasons. For girls, the top three available data in most surveys are insufficient for providing a reasons were pregnancies, marriages, and cost74 (noting quantitative assessment of the potential impact of VIAS on that violence could be one of the reasons leading to the likelihood of dropping out. pregnancies and marriage). Note also that in the PASEC student assessment, “lack of security” is in the top 5 COMPLETION RATES AND EDUCATIONAL reasons for student dropout according to teachers. ATTAINMENT These estimates for Egypt and Uganda with questions asked VACS DATA to youth or teachers and principals as opposed to parents yield estimates of the role of VIAS in drop-outs of the same Another way to look at the potential impact of VIAS is order of magnitude (or smaller in the case of Egypt) as those through surveys that have measures of both VIAS and from parents in the Young Lives Surveys. This general order educational attainment, such as the violence against children of magnitude for the potential share of students dropping surveys (VACS). As mentioned in previous chapters, the out due to VIAS is not really a surprising result – apart data are publicly available for eight countries, so these are the from VIAS, many other reasons are likely to play a more datasets used here (Cambodia, El Salvador, Haiti, Honduras, 74 Wodon et al. (2016). 75 Perlman at al. (2018). 76 Gemignani and Wodon (2017). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 26 Kenya, Malawi, Nigeria, and Tanzania). The surveys cover seems stronger for primary than secondary education, all forms of violence, not only at school, but also elsewhere, possibly because in the countries where a VACS is available, to these are the variables for which potential impacts a smaller share of students complete the secondary level and are estimated77. The focus here is on the probabilities of those students tend to be from more privileged backgrounds completing primary and secondary school, the probabilities and may be less exposed to VIAS. In the case of physical of attending school, and educational attainment as measured violence with injury, the expected years of schooling through years of schooling. The data covers children and completed are reduced by almost four-tenths of a year. For youth aged 13-24 and various age brackets are used for emotional violence, the reduction is at less than one tenth of the various outcomes. Given the time needed to complete a year. Physical and sexual violence are also associated with various cycles, for primary school the focus is on children a reduction in the likelihood of attending school, whether aged 13-17, while for the completion of secondary school, for the full sample or for the sample of children aged 13-17. the focus is on youth aged 18-24. As always, there results should be considered as tentative given limitations of the data and the fact that they represent Key results are provided in Table 2.12. A distinction is made associations and possibly not causal affects. between physical violence leading to injuries, and physical violence not leading to an injury. Note that all locales for Table 2.13 provides simulations of potential benefits from violence are included (school, family, and community). The ending violence against children. Note again that violence interpretation of the results is again in percentage points in all locales is considered. If violence were eliminated, for most indicators, although in the case of the years of the share of students completing primary school could schooling, the interpretation is in terms of the number of increase by 1.1 percentage point, a proportional gain of years completed. For example, a child victim of physical 2.8 percent from the base. When considering younger violence leading to an injury may be 5.9 percentage points children however (those aged 13-18), the potential impact less likely to complete primary school. The potential impact is larger both in absolute and proportional terms, in part of physical violence without an injury on primary school because the base is smaller, and because children affected completion is a bit smaller at 5.9 percentage points. When by violence may not have completed the cycle yet if they potential impacts are statistically significant, they are all repeated a grade. For the likelihood of attending school, suggesting worse educational outcomes. the effects of ending violence against children are of the same order of magnitude. For secondary completion, The evidence of potential impact seems stronger for physical simulations are not conducted since the marginal effects in violence, especially when leading to an injury, than for other Table 3.14 are not statistically significant. Finally, in terms types of violence, although emotional violence is associated of years of schooling, the simulations suggest a gain from with a reduction in years of schooling. The evidence also the elimination of violence against children, but this gain is Table 2.12: Potential Marginal Impacts of Violence by Type on Education Outcomes, VACS Physical with Physical without Emotional Sexual Injury Injury Completion of Cycle Primary Completion (Full sample) -0.059 -0.021 NS NA Primary Completion (Age 13-17) NS -0.049 NS NA Secondary Completion (Age 18-24) NS NS NS NS Currently Attending Attending (Full sample) -0.068 NS NS -0.105 Attending (Age 13-17) -0.109 NS NS -0.080 Educational Attainment Years of schooling -0.387 NS -0.076 NS Source: Authors’ estimations using VACS data. Note: Violence may take place in or around schools, as well as in the family or in the community. 77 This is done by pooling the data for the eight countries together, and controlling for other factors that may affect violence against children. Unfortunately, relatively few such controls are available in the VACS data, so the risk of omitted variable bias cannot be avoided as in other regressions in this study. 27 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 fairly small, especially considering that all forms of violence achievement for individuals in the country, and (3) a are included. Still the suggested effect from the regression combined measure in terms of the learning-adjusted years analysis is positive, confirming in a way the analysis of the of schooling completed by a typical student in each country reasons to drop out of school. (the measure factors in not only the expected number of years of schooling a child is expected to complete, but also CROSS-COUNTRY REGRESSIONS how much the child is expected to learn while in school). Results are tentative given the limitations of the data and As one more test of the potential impacts of VIAS on the fact that cross-country regressions themselves suffer educational attainment, a cross-country dataset was created from various limitations, especially when few controls are with measures of VIAS obtained at the national level from included as is the case here (controls include the dependency the GSHS and HBSC datasets. These measures were ratio, the rate of urbanization, regions of the world, the adult combined with other data at the national level from the literacy rate, and public expenditure for education as a share World Bank’s World Development Indicators, including of GDP). Table 2.14 suggests that countries with higher measures of (1) the expected years of schooling that shares of students involved in fights or bullied tend to do individuals are likely to achieve; (2) the expected learning slightly less well especially on test scores. Table 2.13: Potential Marginal Impacts of Violence by Type on Education Outcomes, VACS Predicted Share Absolute Predicted Proportional Predicted under baseline conditions Change in Share Change in Share Completion of Cycle Primary Completion (Full sample) 80.71 1.10 2.76 Primary Completion (Age 13-17) 68.47 1.66 8.08 Secondary Completion NA NA NA Currently Attending Attending (Full sample) 54.02 1.82 5.64 Attending (Age 13-17) 78.12 1.55 3.59 Educational Attainment Years of schooling 6.87 0.02 0.41 Source: Authors’ estimations using VACS data. Note: Violence may take place in or around schools, as well as in the family or in the community. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 28 Table 2.14: Cross-country Regression for the Potential Impact of VIAS on Educational Outcomes Fewer controls More controls Expected Years Adjusted Years Expected Years Adjusted Years Test Scores Test Scores of Schooling of Schooling of Schooling of Schooling Fights NS -0.005 -0.008 NS -0.005 NS Injuries NS NS NS NS NS NS Bullying 0.002 NS NS NS NS NS Source: Authors’ estimations using VACS data. INDIRECT EFFECTS THROUGH EDUCATIONAL education - on earnings. Apart from educational attainment ATTAINMENT78 and experience, the models may also control for other variables that may affect earnings. The focus in this section is on the indirect potential impact of VIAS through educational attainment on earnings, labor For this study, estimates of the potential impact of education force participation, and perceptions of standards of living. on earnings are based on wage regressions estimated using The previous analysis suggests that VIAS has negative direct the GLD database80 . The aim is to assess the potential impact potential impacts on schooling and learning. How large the on earnings of both educational attainment and learning impacts are remains open to debate, but the fact that there – with literacy as a proxy for learning. Specifically, for are potential impacts is clear. In addition, through both individuals with a primary education or less, a distinction is schooling and learning, VIAS then has an indirect potential made between individuals reporting being literate and those impact on labor force participation and earnings for children declaring not to be. To test for differences in the returns to as they become adults, as well as other outcomes. This is education by gender, models are estimated separately for the focus for this section. The analysis focuses on some of men and women. Some models include only experience and the benefits of higher educational attainment. How these its squared value as additional correlates, while other models benefits can in turn be used to estimate the potential cost include more controls. of inaction for not preventing VIAS will be discussed in subsequent chapters based on estimates of the reduction in Table 2.15 provides the main results in terms of average educational attainment, and especially in the likelihood of coefficient estimates across countries and surveys. As not completing secondary school due to VIAS. Impacts on expected, gains in earnings associated with educational learning will also be considered. attainment are substantial, especially at the secondary and tertiary levels. They are much lower at the primary level. EXPECTED EARNINGS For example, for women with primary education (partial or completed) who are literate, the average expected gain in By affecting children’s educational attainment, VIAS earnings versus no education and no literacy (the reference may affect their productivity in adulthood. Consider first category) is 39.8 percent. By contrast, for women with the benefits from educational attainment for earnings. secondary education, the average gain compared to no There is a large body of literature on this topic79 with the schooling and no literacy is much larger at 129.1 percent. benefits from educational attainment typically measured For women with tertiary education, the average gain is through regression analysis whereby the potential impacts 386.7 percent. on earnings of educational attainment and experience (measured as age minus six and the number of years of Clearly, men and women with higher educational attainment schooling) are estimated. In some models, the focus is on the and literacy in the case of primary education earn more than correlation between years of schooling and earnings, and the those with no education and no literacy, but the gains start implicit gain associated with each additional year of schooling. to be much larger at the secondary and tertiary levels. Also Other models look at the potential impact of different levels important is the quality of the education received, as proxied of schooling, such as having a primary, secondary, or tertiary by whether individuals declare being literate or not. When 78 This section is adapted from Wodon et al. (2019). 79 See Psacharopoulos and Patrinos (2018) for a review. 80 Montenegro and Wodon (2019). 29 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 individuals have a primary education but are not literate, the large as when they have a primary education and are literate. gains are virtually non-existent versus having no education In other words, going to primary school without learning at all. When individuals have a primary education and are does not generate gains, which helps to emphasize the need literate, the gains are larger, and when individuals are literate for learning in school apart from simply going to school or but do not have a primary education, the gains are almost as completing a cycle. Table 2.15: Potential Impacts of Education on Earnings for Men and Women (Percentage Gains, %) Full Sample Latest Year per Country Men Women Men Women No education and illiterate Ref. Ref. Ref. Ref. No education and literate 21.9 30.3 15.8 37.4 Primary education and illiterate 3.6 7.8 6.7 -2.8 Primary education and literate 35.1 39.8 23.6 33.6 Secondary education 101.9 129.1 75.8 108.4 Tertiary education 301.7 386.7 233.2 350.0 Source: Authors’ estimation using GLD database. Note: Reported estimates based on the average value of regression coefficients across counties. The exponential transformation (given that the dependent variable is the logarithm of earnings) is taken for the average coefficient. LABOR FORCE PARTICIPATION Table 2.16 provides results on the potential effects of educational attainment on labor force participation. For Through its impact on educational attainment, VIAS may women, for example, having a secondary education level also affect labor force participation in adulthood. Indeed, is associated with a 9.6 percentage points increase in the apart from leading to higher expected earnings for working likelihood of working versus primary education or less. men and women, a higher level of educational attainment With tertiary education, the potential effect on labor force may also increase their labor force participation or the participation is an even larger gain at the margin of 25.4 number of hours that they work, especially for women. percentage points in comparison to a primary education When women are better educated, the opportunity cost or less. As women with higher levels of education are more of not working or only working part time increases, which likely to enter the labor force, this may result in increases may lead more women to enter the labor force, or work full in the likelihood of working full time, working part-time, or time instead of part time. To measure the potential effect being unemployed. In terms of type of employment, the of educational attainment on labor force participation, the largest increase at the margin from more education is for full- analysis relies on data from the Gallup World Poll for many time work. There is also an increase for part time work and countries. The data allows analysis of the potential impact of unemployment, but to a lower extent and this is not always educational attainment on employment status. statistically significant. Table 2.16: Potential Impact of Educational Attainment on Labor Force Participation Men Women Secondary Tertiary Secondary Tertiary (vs. Primary) (vs. Primary) (vs. Primary) (vs. Primary) Labor force participation 0.02 0.08 0.096 0.254 Working full-time 0.05 0.13 0.090 0.256 Working part-time -0.02 -0.02 NS 0.005 Being unemployed 0.00 -0.02 0.008 NS Source: Authors’ estimation using Gallup World Poll. Note: Regression estimates reported for the pooled sample that includes data for more than 100 countries. NS means that an estimate is not statistically significant at the 10 percent level. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 30 STANDARDS OF LIVING potential impact of educational attainment on monetary poverty, but impacts are likely to be large81 . Using data from Finally, through its effect on educational attainment, VIAS the Gallup World Poll, the potential impact of educational may affect standards of living in adulthood. By increasing attainment on perceptions of standards of living can be earnings and labor force participation for men and women assessed. Results are provided in Table 2.17. Consider again in adulthood, higher levels of educational attainment the case of women. When they have a secondary education contributes to poverty reduction in several ways. Poverty level, they are seven percentage points more likely to state is usually measured by comparing a household’s level of that they have enough money to buy food in comparison income or consumption per capita (or per equivalent adult) to women who have only a primary education or less. With with a poverty line that captures the resources needed by tertiary education, the potential effect for the perceived households to meet their basic needs. The most important ability to satisfy food needs is a gain at the margin of 12 pathways for potential impact are therefore likely to be percentage points in comparison to a primary education or related to (1) higher earnings and consumption for women less, even though individuals with higher levels of educational and their household; and (2) a reduction in household attainment have on average higher expectations for their size and household needs through lower fertility. Higher own standards of living. All measured potential impacts educational attainment for men and women helps not of secondary or tertiary education in comparison to only by increasing the numerator (higher income or lower levels of education in Table 3.19 are positive and consumption), but also by reducing the denominator (smaller statistically significant. households). This study does not provide measures of the Table 2.17: Potential Impact of Educational Attainment on Women’s Perceptions of Standard of Living Men Women Secondary Tertiary Secondary Tertiary (vs. Primary) (vs. Primary) (vs. Primary) (vs. Primary) Not enough money for food 0.02 0.08 -0.07 -0.12 Not enough money for shelter 0.02 0.05 -0.03 -0.06 Satisfied with standard of living 0.004 0.02 0.02 0.07 Source: Authors’ estimation using Gallup World Poll. Note: Regression estimates reported for the pooled sample that includes data for more than 100 countries. NS means that an estimate is not statistically significant at the 10 percent level. 81 UNESCO (2017). 31 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 CHAPTER 3 evidence of potential impacts in the following areas: (1) Health, well-being, and nutrition; (2) Fertility and population MULTI-PRONGED NEGATIVE growth; (3) Decision-making, agency, and other impacts; and (4) Perpetration of violence. It is again important to IMPACTS ON CHILDREN emphasize that violence in schools can have both direct and indirect impacts on children. While in Chapter 4, BEYOND EDUCATION AND most potential impacts of VIAS were estimated directly, in this chapter, after considering estimates of the potential EARNINGS impacts of VIAS on student’s well-being (including in terms of mental health) and risky behaviors from school health surveys, due to data limitations the focus is primarily on the Students who are repeatedly victimized by violence or indirect potential impacts of VIAS through its effect on child bullying may suffer from a broad range of emotional and marriage, early childbearing, and educational attainment. behavioral problems, including sleep disturbances, separation anxiety, hyper-vigilance, physical complaints, irritability, INJURIES regression, emotional withdrawal, blunted emotions and distractibility. Inevitably, bullying and violence have Physical violence especially can lead to injuries, which may consequences lasting well beyond the formative school years. have important health consequences for students and For example, students who are bullied are more likely to healthcare costs for parents. The two main datasets to engage in risky behaviors, such as drinking at an early age, estimate the risk of injuries from VIAS in schools are the using drugs, and engaging early in sexual activities. Violence GSHS data mostly for developing countries and the HBSC also affects students who witness it even if they are not direct for Europe, Canada, and a few countries from the Middle victims. Although bystanders to school violence and bullying East and North Africa. The datasets do not specifically may not feel the direct impacts of it, they can experience identify injuries from VIAS, but in both cases, questions are lasting psychological and emotional scars. Research shows asked about circumstances for injuries. The categories most that student bystanders often feel harm through vicarious likely to reflect at least in part VIAS are “I was attacked or injury, which may trigger personal insecurities and contribute abused or was fighting with someone” for the GSHS, and to trauma, shame or guilt from not intervening to stop an “Fighting” for the HBSC. Data on the share of students attack (especially for students who are highly empathic), fear involved in fights in schools, the share of students injured and anxiety that they may be victims of violence or bullying (all reasons included), and the share injured through fights next, and concern about participating in violence and bullying are provided in Table 3.1. As discussed in previous chapters, resulting from peer pressure from friends. VIAS also has fights are substantial for both boys and girls in both surveys, negative inter-generational effects. Lack of education for suggesting a high level of aggression and correlated issues in mothers is correlated with higher fertility rates, poor health terms of pacific coexistence in schools and school climate. outcomes for their children, and household poverty. Injuries from fights are, as expected, much more likely for boys than girls. The focus in this chapter is on impacts for students directly victimized, versus those witnessing violence, and includes the inter-generational effects of violence. This chapter provides Table 3.1: Injuries from Fights, GSHS and HBSC Data GSHS Data HBSC Data Fought Injured Injured from Fight Fought Injured Injured from Fight Boys 37.4 37.4 2.3 45.3 49.1 2.6 Girls 17.1 24.9 0.6 18.5 40.1 0.9 Total 27.6 31.3 1.5 31.6 44.5 1.8 Source: Authors’ estimations using GSHS and HBSC data. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 32 RISKY BEHAVIORS AND PERCEPTIONS OF on the likelihood of having ever used cannabis. Similarly HEALTH AND WELL-BEING for the HBSC estimates, only one effect is statistically not significant. In the case of PISA data, more effects are GLOBAL ANALYSIS statistically not significant, but most effects are statistically significant with the expected sign. As violence is the result of the accumulation of context-based risk factors, that in turn can be reinforced and perpetuated as What might be the aggregate potential impact on these a consequence to violence, thus feeding a self-perpetuating various outcomes if VIAS were completely eliminated (for vicious cycle, it is important to highlight and critically analyze injuries, only those related to fights are included in the impacts of VIAS on risky behaviors, health, and wellbeing. The simulations since not all injuries are related to VIAS)? As analysis conducted in this section with GSHS, HBSC, and before, the potential benefits from ending VIAS depend PISA data is similar in approach to what was done for schooling on both the sign and magnitude of the marginal impacts and learning in the previous chapter. Data are available for a estimated through the regression analysis, and the share of wide range of indicators. For example, for the GSHS data, students experiencing each form of VIAS. Table 3.3 provides information is available on perceived health (difficulty to the results. The first two columns provide the baseline sleep), risky behaviors (ever smoked, ever used alcohol, ever predicted values of the various indicators, considering first a used drug, ever has sex), and psychological well-being (ever simple average across countries without country population considered suicide, ever planned to commit suicide, ever weights, and next an average accounting for the difference attempted to commit suicide). The indicators available in the between countries in the size of their student body. Consider HBSC data also cover perceived health (self-assessment as one example the likelihood of having difficulties sleeping of health, having headaches, having stomach-ache, having in the GSHS data. With student population weights by back-ache, difficulty to sleep), risky behaviors (ever had sex, country, just over a third of students (37.1 percent) declare ever smoked, ever drank alcohol, ever used cannabis) and having difficulties to sleep. This could potentially be reduced psychological well-being (feeling low, feeling irritable, feeling by 6.9 percentage points by eliminating VIAS according to nervous, and feeling dizzy). In some cases, multiple answers the regressions, or 16.4 percent from the base, which is large. can be provided by students, for example ranging from Other results in the Table can be interpreted in a similar way, strongly agree to strongly disagree, but variables have been suggesting potentially large impacts of VIAS on outcomes dichotomized (yes/no coding) to facilitate interpretation of with all three datasets. the results. The same controls are used for the analysis of the potential impacts of VIAS as in the previous chapter. Detailed data on perceptions of health and well-being as well as risky behaviors are also available in the HBSC and PISA datasets. Table 3.2 provides the key results. Virtually all the potential impacts are again statistically significant, which suggests systematic negative potential impacts. Note that for self-assessment of health, the sign of marginal impacts is negative, which indicates a worse self-assessment. As in the previous chapter, the interpretation of the coefficients is in percentage points. For example, for the likelihood of having difficulties to sleep in the GHSH data, the coefficient 0.058 for physical fights suggests that controlling for other factors that may affect this perception, being involved in a physical fight is associated with an increase in the likelihood of having difficulties to sleep of 5.8 percentage point. With the GSHS data, only one effect is not statistically significant (indicated by NS in the table), for the potential impact of bullying 33 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Table 3.2: Marginal Potential Impacts of VIAS on Various Outcomes Physical Attack Physical Fight Injured Bullied GSHS Data Perceived health Difficulty to sleep 0.058 0.037 0.109 0.125 Risky behaviors Ever smoked 0.051 0.118 0.055 0.065 Ever used alcohol 0.049 0.139 0.069 0.078 Ever used drug 0.021 0.042 0.031 0.018 Ever has sex 0.031 0.097 0.038 0.027 Psychological well-being Ever considered suicide 0.044 0.036 0.049 0.072 Ever planned to suicide 0.039 0.035 0.041 0.06 Ever attempted to suicide 0.068 0.038 0.061 0.069 HBSC Data Perceived health Good self-assessment of health - -0.028 -0.020 -0.043 Having headaches - 0.071 0.099 0.079 Having stomach-ache - 0.077 0.093 0.080 Having back-ache - 0.078 0.091 0.069 Difficulty to sleep - 0.068 0.077 0.079 Risky behaviors Ever had sex - 0.165 0.045 -0.000 Ever smoked - 0.129 0.033 0.017 Ever drank alcohol - 0.181 0.056 0.024 Ever used cannabis - 0.109 0.032 NS Psychological well-being Feeling low - 0.085 0.051 0.142 Feeling irritable - 0.102 0.058 0.079 Feeling nervous - 0.080 0.051 0.094 Feeling dizzy - 0.074 0.106 0.086 PISA Data Perceived health Difficulty to sleep NS 0.022 NS 0.089 Poor self-assessment of health 0.004 NS NS 0.004 Having headaches -0.028 NS -0.018 0.079 Having stomach-ache NS 0.034 NS 0.072 Having back-ache NS 0.035 0.021 0.062 Psychological well-being Feeling depressed NS 0.015 NS 0.124 Feeling irritable -0.046 NS NS 0.095 Feeling nervous -0.041 0.017 NS 0.073 Feeling dizzy 0.026 0.042 0.018 0.096 Feeling anxious NS 0.021 NS 0.111 Source: Authors’ estimations using GSHS and HBSC data. Note: NS means statistically not significant at 0.1 level. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 34 Table 3.3: Simulations of Potential Impacts of Ending VIAS on Perceptions of Well-being, GSHS/HBSC Predicted Share Under Potential Impact of VIAS Share Associated with VIAS Baseline Conditions Simple Weighted Simple Weighted Simple Weighted Average Average Average Average Average Average GSHS Data Perceived health Difficulty to sleep 37.1 31.0 6.9 5.2 16.4 13.7 Risky behaviors Ever smoked 16.6 19.0 5.8 5.2 28.5 21.8 Ever used alcohol 31.4 24.1 6.5 5.0 21.4 18.2 Ever used drug 8.9 10.6 3.7 3.4 31.9 24.5 Ever has sex 10.8 9.7 3.6 3.0 28.2 22.5 Psychological well-being Ever considered suicide 13.7 13.0 4.7 3.5 27.3 21.8 Ever planned to suicide 11.1 9.2 3.8 2.8 27.1 21.8 Ever attempted suicide (*) 9.3 8.6 4.2 3.2 32.9 26.3 HBSC Data Perceived health Poor self-assessment of health 13.5 14.5 2.1 2.3 14.0 14.0 Having headaches 52.6 53.7 4.2 4.3 7.7 7.7 Having stomach-ache 47.9 49.5 4.3 4.4 8.8 8.7 Having back-ache 40.1 41.0 4.1 4.2 9.6 9.6 Difficulty to sleep 48.6 50.0 4.1 4.2 7.8 7.8 Risky behaviors Ever had sex 10.4 10.2 3.1 3.1 19.3 19.4 Ever smoked 17.5 18.3 4.3 4.5 20.9 20.8 Ever drank alcohol 38.6 39.0 5.2 5.2 13.0 13.0 Ever used cannabis 6.6 6.8 2.0 2.1 21.2 21.0 Psychological well-being Feeling low 50.3 51.9 5.9 6.0 11.0 11.0 Feeling irritable 66.1 67.9 5.1 5.1 7.5 7.3 Feeling nervous 62.2 63.6 4.8 4.9 7.4 7.4 Feeling dizzy 31.9 32.9 4.5 4.6 12.4 12.4 PISA Data Perceived health Difficulty to sleep 48.8 49.9 3.4 3.5 5.6 5.9 Poor self-assessment of health 0.6 0.7 0.2 0.2 17.3 18.4 Having headaches 60.3 60.8 1.5 1.6 2.2 2.3 Having stomach-ache 54.4 55.1 3.1 3.3 5.2 5.4 Having back-ache 55.6 56.3 3.2 3.3 5.0 5.2 Psychological well-being Feeling depressed 59.8 60.4 3.1 3.3 4.4 4.6 Feeling irritable 57.5 58.4 1.9 2.0 2.8 2.9 Feeling nervous 56.2 57.1 1.9 2.0 3.0 3.1 Feeling dizzy 33.4 34.6 4.6 4.9 10.3 10.7 Feeling anxious 52.2 53.0 3.8 4.0 6.0 6.3 Source: Authors’ estimations using GSHS and HBSC data. Note: The simple average is an average values across all countries without considering differences in student population between countries. The weighted average factors in the countries’ student population. (*) The proportion of children attempting suicide is high and there may be an issue of data quality (e.g., some children saying they attempted suicide also say they did not consider or plan a suicide). 35 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 ANALYSIS WITH VACS DATA women; Men need to have sex with other women, even if Self- perceptions of the health and well-being of children they have good relationships with their wives; Women who and youth can also be assessed with VACS. As previously carry condoms have sex with a lot of men; A woman should mentioned, data are publicly available for eight countries tolerate violence to keep her family together; and finally, A (Cambodia, El Salvador, Haiti, Honduras, Kenya, Malawi, woman should not complain to anyone when she is beaten Nigeria, and Tanzania) and these datasets are used for this or otherwise abused by her man in order to keep the peace analysis . The surveys cover violence at school and in the in the family. Analysis is conducted with both dichotomic family and community. The analysis is performed by merging variables combining the categories and an overall index for the datasets and exploring correlates of a dozen indicators. both questions. Similar results are obtained. The focus is on the potential impacts of physical, emotional, and sexual violence controlling for other factors that could Key results are provided in Table 3.4. As in the previous affect those indicators (relatively few controls are available in chapter, episodes of violence in all locales are considered, not the VACS data, so there is a risk of omitted variable bias). only those taking place in and around schools. A distinction is made as to whether physical violence leads to injuries or Outcomes are considered in four areas: Friends and attitudes not. The interpretation of the results is in percentage points. (whether individuals often talking to friends about important For example, having been the victim of emotional violence things, and whether they do not tolerate attitudes that are is associated with a reduction in the likelihood of often detrimental to women in terms of both wife beating and talking to friends about important things of 2.6 percentage sexual practices); Sexual and reproductive health (whether points. As another example, having been the victim of girls got married before the age of 18, had a first child sexual violence reduced by 2.0 percentage points the lack of before the age of 18, or had sex before the age of 18); Trust tolerance towards wife beating, which means that tolerance (whether individuals trust people in neighborhood, and towards wife beating increases. When potential impacts whether they feel close to their biological mother and father); are statistically significant, they lead to worse outcomes, and Health and well-being (whether individuals feel bad about with few exceptions. For child marriage, the effects are their own health, whether they had suicidal thoughts, and also statistically significant, even though they are small. whether they feeling safe in their neighborhood). The largest impact is for emotional violence and feelings about one’s health. The second largest potential impact On tolerance towards wife beating, the question is whether is for the relationship between sexual violence and having it is right for a man to hit or beat his wife if she goes out sex before the age of 18, but this may reflect the fact that without telling him, if she does not take care of the children, sexual violence may be forced sex. Other potential impacts if she argues with him, if she refuses to have sex with him, of violence on the outcomes that are near or above 10 or if she burns the food. As to the tolerance question percentage points are for indicators related to feelings about related to sex, the questions are whether men, not women, one’s health and suicidal thoughts. This is observed for both should decide when to have sex; Men need more sex than physical violence with an injury and sexual violence. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 36 Table 3.4: Potential Marginal Impacts of Violence (All Locations) on Various Outcomes, VACS Physical Violence Physical Violence Emotional Sexual with Injury without Injury Violence Violence Friends and Attitudes Talking to friends (important things) NS NS -0.026 NS Lack of tolerance for wife beating NS 0.012 -0.009 -0.020 Lack of tolerance for sexual practices NS NS NS -0.010 Sexual and Reproductive Health Child marriage NS 0.000 0.000 0.000 Child pregnancy 0.043 NS 0.017 0.014 Having sex before 18 NS NS 0.063 0.145 Trust Trust people in neighborhood NS -0.032 -0.106 -0.081 Feeling close to biological mother NS -0.021 -0.052 -0.016 Feeling close to biological father -0.077 -0.034 -0.084 -0.042 Health and well-being Feel bad about one’s health 0.125 0.065 0.177 0.099 Suicidal thoughts 0.104 0.028 0.095 0.056 Feeling safe in neighborhood NS NS -0.048 -0.021 Source: Authors’ estimations using VACS data. Note: Violence may take place in or around schools, as well as in the family or in the community. Table 3.5 provides simulations of potential benefits from thoughts having sex before the age of 18 (reduction by 4.9 ending violence against children. If all forms of violence points), and feeling close to one’s biological father (increase were eliminated, some of the largest effects in percentage by 4.9 points). In proportional terms from the base, however, points changes are observed for the probabilities of feeling the largest effects are observed for suicidal thoughts and for bad about one’s health (reduction of 6.1 percentage points), early childbearing (having a child before age 18)82. trusting other people (increase by 6.0 points), having suicidal 82 Proportional changes are not equal in the Table to the absolute changes divided by the baseline values because simulations are conducted for each country, then aggregated (the mean of a ratio is not the ratio of the means). 37 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Table 3.5: Simulations of Potential Impacts of Ending VIAS on Various Outcomes, VACS Predicted Share Under Absolute Predicted Change in Proportional Predicted Change Baseline Conditions Share with no VIAS in Share with no VIAS (*) Friends and Attitudes Talking to friends (important things) 67.43 1.07 1.91 Lack of tolerance for wife beating 88.32 0.78 1.55 Lack of tolerance in other areas 93.09 0.39 0.45 Sexual and Reproductive Health Child marriage 20.43 -1.38 -5.37 Child pregnancy 9.60 0.81 28.90 Having sex before 18 38.04 -4.90 -12.88 Trust Trust people in neighborhood 71.54 6.03 10.96 Feeling close to biological mother 92.81 2.54 2.97 Feeling close to biological father 78.28 4.84 7.21 Health and well-being Feel bad about one’s health 45.62 -6.09 -15.61 Suicidal thoughts 8.98 -5.26 -33.53 Feeling safe in neighborhood 84.98 2.87 5.14 Source: Authors’ estimations using VACS data. Note: (*) The proportional change is not the absolute change divided by the baseline because analysis is conducted for each country, then aggregated (the mean of a ratio is not necessarily the ratio of the means). INDIRECT EFFECTS THROUGH CHILD MARRIAGE, EARLY CHILDBEARING, FERTILITY, AND EDUCATIONAL ATTAINMENT POPULATION GROWTH The above analysis using a range of different surveys There is a strong mutual relationship between girls’ education suggests that VIAS has negative direct potential impacts and child marriage, defined as a girl entering in a formal or on health and well-being. It may also lead to an increase in informal union before the age of 1882. Especially in countries the probability of having sex before the age of 18, marry where the prevalence of child marriage is high, parents often before the age of 18 (although the effect is small), and have have their daughters marry early when they are not in school, a first child before the age of 18. In addition, as discussed in part, because of a concern that they may otherwise in Chapter 2, VIAS leads to lower educational attainment, engage in sexual activity. In many contexts, a pregnancy leading students to drop out of school prematurely. For girls outside of marriage leads to ostracism, thereby affecting especially, this in turn may increase in many countries the a girl’s prospects in life. As VIAS reduces educational likelihood that they will marry or have children as children. attainment for girls, it also contributes indirectly to child How large the impacts of VIAS are on various outcomes marriage. Child marriage, in turn, is a key driver of teen can be debated, but the fact that there are wide-ranging pregnancies and early childbearing (having a child before potential negative impacts is clear. This leads to a wide range the age of 18). There are differences between countries in of negative indirect potential impacts of VIAS on other the strength of those relationships. Still, for most developing indicators. This section considers these indirect impacts in countries, most women having a child before the age of 18 do four domains: (1) Fertility and population growth, including so probably because of child marriage85 . a discussion on child marriage and early childbearing since they are key drivers of fertility rates especially in low income Analysis of DHS data suggests that keeping girls in school is countries; (2) Health and nutrition; (3) Decision-making one of the best ways to end child marriage and reduce the and agency; and (4) Perpetration of violence. The analysis is risk of early childbearing. There is also a strong relationship adapted from a study at the World Bank on the cost of not between girls’ educational attainment, child marriage, and educating girls83 . total (lifetime) fertility. Both low educational attainment and 83 Wodon et al. (2018). 84 Field and Ambrus (2008); Nguyen and Wodon (2014). 85 Wodon, Male, and Onagoruwa (2020). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 38 child marriage lead women to have children earlier in life, understanding of what they need to do to care for a child and more children over their lifetime. The analysis considers when sick or injured. As low education attainment for the number of children that women have towards the end of mothers increases their risk of exposure to IPV, it may have their reproductive age86 . Analysis for developing countries spillover effects for children. Toxic stress90 responses for based on DHS data suggests that having completed children can have damaging effects on learning, behavior, secondary education is almost always associated with a and health. Even when children are exposed to IPV in reduction in total fertility in comparison to no education or utero, they tend to have worst health at birth and increased incomplete primary. In addition, because universal secondary mortality rates. completion could virtually eliminate child marriage, this would lead to an additional reduction in total fertility rates. Overall, Consider in particular the issues of under-five mortality and the analysis suggests that universal secondary education could stunting. Stunting often results from persistent insufficient lead to a reduction in total fertility of up to a third in two nutrient intake and infections and may lead to delayed motor dozen countries for which analysis was undertaken87. development and poor cognitive skills later in life. Estimates of the potential impacts of a mother’s education level on the Through its potential impact on total fertility, low educational risks of under-five mortality and stunting suggest that gains attainment for girls and child marriage contribute to higher could be achieved from higher educational attainment for population growth, which weakens the ability of governments mothers. The same is true for birth registration as a higher to provide basic services of sufficient quality to the level of educational attainment for mothers is positively population. It also reduces the ability of countries to benefit correlated with the likelihood of registering their child from the demographic dividend, with major implications for at birth. These are just some of the beneficial effects of the population’s standards of living. educational attainment – many others effects have been documented, including for agency and decision-making HEALTH, NUTRITION, AND AGENCY ability in a wide range of areas. A lack of education may affect women’s health as they may be less aware of how to take care of themselves when sick or injured. It may also lead to lack of knowledge about STDs. And through its impact on child marriage and early childbearing, a lack of education may lead girls to give birth at a young age, which increases risks of maternal mortality and morbidity88 . Other risks related to low educational attainment include malnutrition, isolation, depression, an inability to negotiate sexual and reproductive behaviors, higher risks of suffering from IPV, and a lower ability for women to make decisions related to their own healthcare. Consider next children. Early childhood is critical for a child’s development89. Risk factors early in life may affect brain development and capabilities, with lasting consequences in adulthood, including the ability to earn a decent wage. A lack of educational attainment for mothers is associated with worse health outcomes among their children. This may be because better educated mothers have a better 86 Onagoruwa and Wodon (2018). The term “total fertility” is defined here as the number of live births that a woman has over her lifetime. This definition is used for individual-level econometric work to measure the marginal impact of child marriage on fertility. By contrast traditional “total fertility rates” are population-level estimates. 87 Part of the potential effect of educational attainment and child marriage on total fertility may come from the use of modern contraceptive methods since such use tends to increase with higher educational attainment and when women do not marry as children, at least in some countries. 88 Nove et al. (2014). 89 Black et al. (2017). 90 Toxic stress refers to a child’s response when experiencing repeated adversity, including physical and emotional abuse as well as economic hardship and exposure to violence in the household (such as intimate partner violence). 39 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 CHAPTER 4 COUNTRY ESTIMATES COST OF INACTION Selected country estimates are provided in Table 4.1. In the first study91 mentioned in the Table, costs were estimated in a simple way as the product of (1) the lifetime loss in VIAS contributes to learning poverty, low educational earnings from dropping out of school before completing attainment, and poor health, which in turn have major secondary school (expressed as a share of Gross Domestic negative impacts for children, their future household and Product); and (2) an assumed share of drop-outs due to their own children, their communities, and societies as a violence against children, set at one percent or five percent, whole. What are the economic costs associated with those given that many other factors may lead children to drop out impacts? For many impacts, this is a hard question to answer, of school, especially in developing countries. The lifetime but for a few, estimates can be provided. The focus in this loss in earnings from dropping out of school versus earnings chapter is on the higher levels of human capital wealth that that would have been obtained if children had continued could result from better health and educational outcomes if their education were estimated at 14.4 percent of GDP for VIAS were eliminated. The analysis proceeds in three steps. Brazil, 4.6 percent for the United States, and 12.8 percent First, a brief review of the literature on the cost of violence for India. If one percent of drop-outs are due to violence against children is provided. Next, we provide estimates of against children, the cost of violence against children due to the potential cost of VIAS related specifically to its impact lost productivity in adulthood would then be valued at 0.14 on children dropping out of school prematurely or not percent of GDP for Brazil, 0.05 percent for the United learning, and how this may affect earnings in adulthood. States, and 0.13 percent for India. When five percent of Finally, the chapter discusses briefly two other potential drop-outs are assumed to be due to violence against children, economic costs related to lower educational attainment due the estimates are five times larger. Even then, in all three to VIAS – the costs related to population growth and under- countries, costs would be under one percent of GDP. five mortality and stunting. In low income countries, these In some cases costs as a share of GDP are higher (for costs may be large in comparison to prevailing standards of Central American countries), and in other cases lower living, even if at the global level they may not be seen as large (for European countries). because the countries where population growth as well as under-five mortality and stunting are high account for a small share of global human capital wealth. LITERATURE ON THE COST OF VIOLENCE AGAINST CHILDREN Several studies have been conducted to estimate the economic costs of violence against children. Some of these studies are at the country level or focus on a few countries. One would expect that various studies would provide different estimates of the cost of violence against children – since both the prevalence of violence against children in its various forms and its economic costs may vary between countries. But differences in costs also relate to differences in methodology. In particular, some global estimates appear to be on the high side in comparison to estimates at the country level. 91 Pereznieto et al. (2010). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 40 Table 4.1: Selected Country Estimates of the Economic Cost of Violence against Children Countries/regions Outcomes and Costs Types of Violence Cost Share of GDP Selected countries Productivity costs due to lower Violence against Various Less than 1% Pereznieto et al. (2010) educational attainment children Health care costs, productivity United States Nonfatal and fatal US $124 losses, child welfare, criminal 0.84% Fang et al. (2012) child maltreatment billion(2008) justice costs, special education Illicit drug use, smoking, problem East Asia and Pacific drinking, early sex, teenage Violence against US $206 billion 2.0% Fang et al. (2015) pregnancy, self-harm, stomach children (2012) pain, mental disorder Mental disorder, smoking, China Violence against US $101 billion problem drinking, illicit drug use, 1.7% Fang et al. (2015) children (2013) self-harm Health consequences and Cambodia Violence against US $251.3 productivity losses from lower 1.65% Fang (2015) children Million (2013) educational attainment Health costs, special education, criminal justice, housing, child Australia Violence against US $26.7 protection, productivity losses, 1.5% McCarthy et al. (2016) children billion(2012) deadweight losses, impacts on lifespan and quality of life South Africa Health and behavioral outcomes, Violence against US $13.5 billion 4.3% Fang et al. (2017) fatal cases children (2015) Health outcomes and productivi- Nigeria Violence against US $15.0 ty losses due to lower educational 2.7% UNICEF (2019) children billion(2018) attainment Health outcomes, productivity Australia Violence against AU $34.2 billion losses due to lower educational 2.6% Deloitte Access Economics (2019) children & youth billion(2016-17) attainment, justice and housing Source: Compiled by the authors. More detailed studies were conducted in recent years. Many medical costs, child welfare costs, criminal justice costs, and of these studies suggest (lifetime) costs from violence against special education costs). children at about two percent of (annual) GDP on average, with estimates ranging from 0.8 percent for the United The study for Nigeria93 in Table 4.1 estimates that one in two States92 to 4.3 percent for South Africa. The fact that children report being a victim of physical violence before many of these estimates are higher than those based only the age of 18, with the proportions being at one fifth for on losses in earnings makes sense since additional costs are emotional violence and one fourth for sexual violence. The included, but this is not always the case. For example, a study economic costs for health outcomes (lost disability-adjusted for the United States92 puts the baseline economic burden life years or DALYs94) associated with violence are valued of child maltreatment at US$124 billion in 2008 (0.84 at US$8.9 billion (1.6 percent of GDP), while the cost in percent of GDP), although sensitivity analysis suggests lost earnings is valued at US $6.1 billion or 1.1 percent of that costs could be as high as US$585 billion (close to four GDP based on estimates of the marginal effect of violence percent of GDP). The analysis is based on an estimated on secondary school completion and the ensuing loss in cost of US$210,012 per child being maltreated with no earnings. This would yield a total cost of 2.7 percent of GDP. fatal consequences (in 2010 dollars and with a discount rate of three percent). Two thirds of the costs of non-fatal As another example, the last study mentioned in Table 4.1 maltreatment are from losses in productivity, with the rest for New South Wales suggests costs at AU$11.2 billion for related to other costs (childhood health care costs, adult that part of the country, of which AU$2.3 billion are costs 92 Fang et al. (2012). 93 UNICEF (2019). 94 As defined by the WHO, one DALY represents the loss of the equivalent of one year of full health. DALYs for a disease or health condition are the sum of the years of life lost to due to premature mortality (YLLs) and the years lived with a disability (YLDs) due to prevalent cases of the disease or health condition in a population. 41 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 borne by the government, AU$8.2 billion are costs incurring study estimates a total cost of US$9.5 trillion or 11 percent to individuals and communities, and AU$600 million are of global GDP. The cost of homicides is much larger than costs borne out by the federal government. When the the cost of civil conflict, but the largest cost is for violence at estimate is extrapolated for Australia as a whole, the cost of home, the most prevalent type of violence affecting children violence against children in the country is estimated at AU as well as women. The cost of violence against children is $34.2 billion, which would represent about 2.6 percent of estimated at $3.7 trillion, of which $3.6 trillion is related to GDP (the authors do not provide this comparison of lifetime child abuse, defined as being slapped on the face, head or costs with annual GDP, but it is included in Table 5.1 for the ears, and/or beaten repeatedly with an implement. The other purpose of comparing the result with the other estimates in costs for children relate to the number of children being the Table). victims of homicides and reported child sexual violence. These costs are much lower, each at about one percent of the GLOBAL ESTIMATES cost of child abuse, but prevalence may be underreported. Table 4.2 provides summary results for two global studies. The estimate of $3.6 trillion for child abuse is obtained by The first study95 appears to follow the methodology adopted multiplying a cost per episode by the number of episodes. in the first study in Table 4.1, although details on estimations The cost per episode is based on data on the cost of assault are not provided. The study focuses on losses in productivity from the Institute for Economics and Peace97, with the due to lower educational attainment. It puts the cost of authors noting that this cost is of a similar order of magnitude violence against children at up to US$7 trillion globally or to the median cost in a review of the literature and the cost eight percent of global GDP. This estimate is the upper of non-fatal abuse reported for the United States in the bound (highest cost scenario). A lower-cost scenario leads second study in Table 4.198 . The authors scaled up or down to costs valued at two to five percent of global GDP. In at the country level the cost per abuse by GDP per capita comparisons to estimates in Table 4.1, the costs appear since costs observed for the United States would be too to be on the high side, especially since these estimates high for many other countries. The authors estimate the account only for lost productivity in adulthood due to lower cost of child abuse at 4.2 percent of global GDP with major educational attainment, and not for other costs, as is the case differences between regions. The cost at 1.9 percent of GDP for many studies mentioned in Table 4.1. in high income countries appears reasonable in comparison to the country studies in Table 4.1. However, for developing The second study in Table 4.2 attempts to measure the countries, the cost is higher. In sub-Saharan Africa, the costs of multiple forms of violence – including welfare estimate is at 19.9 percent of GDP, which seems unrealistic costs of collective violence as well as costs of interpersonal even if one acknowledges that the prevalence of violence is violence, harsh child discipline, IPV, and sexual abuse. The higher in that particular country. Table 4.2: Selected Global Estimates of the Economic Cost of Violence against Children Countries/regions Outcomes and Costs Types of Violence Cost Share of GDP Global Productivity costs due to lower Violence against US$ 2 trillion to 2% to 8% Pereznieto et al. (2014) educational attainment children US$ 7 trillion Multiple outcomes with costs Domestic child Fearon and Hoeffler (2014) per episode of violence based on US$ 3.6 trillion 4.2% abuse previous studies Source: Compiled by the authors. 95 Pereznieto et al. (2014). 96 Fearon and Hoeffler (2014). 97 Institute for Economics and Peace (2014). 98 Fang et al. (2012). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 42 Table 4.3: Estimates and Components of the Changing Wealth of Nations Total Wealth in 2014 Per Capita Wealth in 2014 (US$ trillions) (US$) Total wealth 1,143.2 168,580 Produced capital 303.5 44,760 Natural capital 107.4 15,841 Human capital 736.9 108,654 Of which men 453.2 66,832 Of which women 283.6 41,823 Net foreign assets -4.6 -676 Source: Lange et al. (2018). OTHER STUDIES of “wasted” public spending for primary education. This calculation is based on GDP in low and lower-middle income Beyond studies providing national or global estimates for countries, the average share of GDP spent on education, health and productivity costs (these are often the two the share of public funding allocated to primary schooling, largest categories of costs), other studies focus on other and the median duration of the primary cycle. The loss of types of costs. For example, a USAID factsheet on the one year of schooling on average due to SRGBV amounts cost of school-related gender-based violence (SRGBV) to US$17 billion in lost public spending for education in low suggests a cost of US$17 billion in low and middle income and middle income countries. The note suggests that the countries99. The estimate comes from the assumed impact estimate is conservative since it only includes children who of SRGBV on learning performance in international student stayed in school and does not include other costs (such as assessments (at about 30 points), which may translate losses in earnings in adulthood). Still, the estimated loss into the equivalent of a loss of one year of schooling due in investments in education is larger than total overseas to lack of learning. Next, the analysis considers what that development assistance for education which is about $13 potential loss of one year of schooling represents in terms billion depending on the year. 99 USAID (2015). 95 This section is adapted in part from Wodon et al. (2018). 43 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 POTENTIAL LOSSES IN HUMAN CAPITAL Estimates of the changing wealth of nations are available WEALTH FROM VIOLENCE IN SCHOOLS100 from the World Bank for 141 countries with a population The costs of violence against children mentioned in global of 6.8 billion people in 2014 (95 percent of the world’s studies appear high in comparison to more detailed estimates population)101 . In those estimates, human capital wealth for specific countries. This may be due to assumptions and represents the expected future earnings of the labor force data limitations. Importantly, estimates of costs as a share of in net present value. As shown in Table 4.3, global wealth GDP do not mean that year after year, that share of GDP is stood at US$ 1,143 trillion in 2014. Human capital wealth necessarily lost due to violence against children, since costs was at US$ 737 trillion, thus accounting for two-thirds over a lifetime are compared to income (GDP) for a single of total wealth. In per capita terms, total wealth stood year. This may not always be clear to readers not familiar with at US$ 168,580 per person, with human capital wealth the methodology. Not all studies are affected by this issue, estimated at US$108,654 per person. Inequality in human as it depends on how estimates are conducted, but some capital and total wealth between countries is high. In high may be. To be consistent in the timeframes being used, it income OECD countries, total wealth per capita is above seems best to measure lifetime costs from violence. To do US$700,000, and human capital wealth is at close to so, we rely on estimates of the wealth of nations, and more US$500,000 per person. This is more than 90 times specifically on human capital wealth. the level in low income countries. Table 4.3 also provides estimates of human capital wealth by gender. Due to gender ESTIMATES OF HUMAN CAPITAL WEALTH inequality in earnings, human capital wealth attributed to women was US$ 283.6 trillion in 2014 versus US$ 453.2 A country’s wealth includes natural capital such as trillion for men102. agricultural land, forest, oil, gas and minerals, to give a few examples. It also includes produced capital – such as LOSS IN HUMAN CAPITAL WEALTH FROM infrastructure, machinery, factories, or buildings. VIOLENCE IN SCHOOLS Finally, a country’s wealth includes human capital, such as a well-educated, healthy, and productive labor force. These To measure potential losses in human capital wealth due three categories – produced, natural, and human capital, to violence in schools, simple simulations are used. First, are considered the three main components of the changing losses in learning while in school translate into losses in the wealth of nations, which together with net foreign assets, education component of human capital wealth. Estimates of provide the assets base that countries rely on to produce losses in learning due to VIAS are computed at the country GDP capita from year to year. level based on the estimate of the impact of VIAS on Table 4.4: Potential Loss in Human Capital Wealth Due to Violence in School Total Cost (US$ trillions) Human capital wealth in 2014 736.9 Losses due to children not learning as much Loss in human capital wealth for countries included in PISA 4.7 Scaled-up losses for all countries with measures of human capital wealth 5.7 Losses due to children dropping out of school Potential gain in human capital wealth with universal secondary education 77.3 Share of drop-outs due to violence in school 5% Potential gain from ending school drop-outs due to violence in school 3.9 Combined estimate of losses Combined losses from lack of learning and children dropping out of school 9.6 Potential gain from ending violence in school as share of human capital wealth 1.3% Scaling-up factor from 2014 to 2019 Cost of VIAS in 2019 (using ratio of global GDP between 2019 and 2014 at 1.1) 10.6 Source: Authors’ estimations. 101 Lange et al. (2018). 102 See Wodon and de la Brière (2018) for more details. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 44 LIMITS OF THE ANALYSIS AND OTHER POTENTIAL COSTS learning as measured in the PISA assessment (see Chapter 2). In so doing, only the component of human capital wealth While the above analysis is illustrative of the fact that the due to education is taken into account103 . The estimates economic costs of VIAS are very large, it should be clear that obtained for all countries with a PIAS assessment are then the estimates consider only costs related to losses in learning scaled up to a global estimate using the ratio of global human and educational attainment from VIAS and not other direct capital wealth to human capital wealth in PISA countries. or indirect costs. Large gaps in in our understanding of the As shown in Table 5.4. This leads to a loss in human capital potential costs of VIAS remain at the global level, for three wealth globally of US$ 5.7 trillion. main reasons: Next, an additional estimate of cost is computed for children 1. Data limitations on the types of VIAS considered in this dropping out of school specifically because of violence in study: Because the analysis in this study relies in large school. The assumption used is that those children may have part on school health surveys and student assessments, been able to pursue their education up to the completion of the focus is mostly on physical and emotional violence secondary school if they had not dropped out. In low income and less on sexual violence even if some analysis is countries, this may be an optimistic assumption, but in conducted with VACS and DHS data on sexual violence. upper-middle and high income countries most of the world’s human capital wealth is concentrated, this may indeed be the 2. Data limitations for costing the direct impacts of VIAS: case. Data are, however, not available at the country level on While various costs of VIAS have been estimated in the share of drop-outs due to VIAS. The assumption used for the literature for specific countries, obtaining global all countries is that five percent of all children dropping out estimates for many types of costs requires data that are do so because of VIAS. In that case, five percent of the loss typically not available for many countries. For example, in human capital wealth due to children not completing their exposure to violence may affect brain development secondary education is assumed to be due to VIAS. in ways that are not accounted for here. As another As shown in Table 4.4, this leads to a loss in wealth of example, medical costs associated with VIAS are also US$3.9 trillion. not included here. The combined loss from the potential impact of VIAS 3. Data limitations on indirect impacts of VIAS and their on educational achievement and attainment is valued at costs: As just one example of indirect impact, the US$9.6 trillion in 2014. Given population growth and gains fact that VIAS has negative effects on educational in standards of living between 2014 and 2019, that value is attainment and learning may have intergenerational scaled up by the ratio of global GDP in current US$ in both effects. Those costs are not included in the above years104 , leading to an estimated loss in human capital wealth analysis, but may nevertheless be substantial, as in 2019 (pre-COVID) of just under US$11 trillion. discussed briefly below for fertility and child health. These estimates are nothing more than broad orders of The analysis in Chapter 3 noted that women’s educational magnitude – they are not meant to be precise or definitive attainment has a large impact on lifetime fertility and given the many assumptions involved. But they appear to population growth. For example, a World Bank study on be of a reasonable order of magnitude when compared to child marriage105 suggests that the world population could country level estimates in Table 4.1, which tend to be more have been 1.4 percent lower in 2030 if child marriage carefully done than global estimates. Indeed, in Table 4.1, could have been ended in 2015 versus business-as-usual the average loss as a share of GDP is of the order of two trends. Universal secondary education could lead to an even percent. In Table 4.4, the loss as a share of human capital is larger reduction in population growth (the effect could be at 1.3 percent. This is smaller but logical since only losses in about three times as large). As ending VIAS would increase earnings are accounted for, as opposed to other costs related educational attainment for girls and thereby reduce child to VIAS. marriage, it would reduce total fertility rates, thus reducing 103 That component is itself estimated using the methodology for the measurement of human capital wealth, but setting the returns to experience to zero in the estimations. 104 Per capita GDP and human capital wealth are very highly correlated, hence the extrapolation makes sense. 105 Wodon et al. (2017). 45 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 population growth. With global wealth at US$1,143 trillion in economic losses equivalent to four to 11 percent of GDP in 2014, even a small annual reduction in population growth can sub-Saharan Africa and Asia106 . Results from an experimental lead to a non-negligible increase in wealth per capita which study in Guatemala suggest that children who avoid stunting might be valued at the global level in the trillions of dollars have better cognitive abilities and higher levels of per capita over a decade or two. consumption in adulthood107. As early childbearing leads to higher risks of stunting for the children of young mothers, A similar argument, albeit with lower valuations, could be by reducing the risk of early childbearing for girls as a result made for the intergenerational impact of ending VIAS on of higher educational attainment, ending VIAS would lead the risk of under-five stunting and mortality for children. to economic benefits as a result of a reduction in under-five It has been suggested that undernutrition may lead to stunting (and mortality) across generations. 106 Horton and Steckel (2013). 107 Hoddinott et al. (2013). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 46 PART II approach as that used for many communicable and non- communicable diseases. Focusing on prevention, the model PROGRAMS AND POLICIES is dynamic and gender-sensitive, and it focuses on the root causes for violence and the accumulation of disadvantages or layers of exclusion in a child’s life that makes him or her more prone to behave violently or experience violence108 . CHAPTER 5 By identifying key risk factors in a given context instead FACTORS LEADING TO of considering violence as a given of human nature and treating violent individuals as abnormal or bad persons, VIOLENCE the model helps us to break the cycle of violence whereby punitive responses may lead to more violence. Depending on each stakeholder’s capacity to act and level of influence, There is growing international evidence on what works to the model helps to find entry points for interventions and prevent VIAS. Several examples of successful interventions mainstream violence prevention programs in public policy will be discussed in Chapters 6 and 7. But first, to provide and education systems. By recognizing shared responsibility a framework for such interventions and understand their for violence prevention, it helps to focus on ways to change logic, it is useful to consider the factors that lead to VIAS. mindsets and behaviors. Risk factors for violence are Preventing violence requires an understanding of the factors typically recognized at the levels of individual, relationship, that influence it. No single factor can perfectly predict community, and society. For this study, because the incidents of violence. Rather, it is often a combination of risk focus is on VIAS, an additional level is included within the factors at the individual, relationship, school, community, and community corresponding to the school. Some of the risk societal levels that leads to violence. This short chapter relies factors for perpetrating violence or experiencing violence are on the ecological model as a conceptual device (as well as a determined by biology, others are environmental, and yet governance tool when implementing programs and policies) others are behavioral (Figure 5.1). to better understand why and where violence occurs, and suggest options for programs and policies to prevent violence • At the individual level, childhood disruptive behavior, through so-called protective factors. These protective factors some of which can take place in schools (opposition, in turn underline the types of interventions and approaches aggression and hyperactivity with corresponding poor to prevent VIAS that are discussed in the next two chapters. self-esteem, poor emotions management, impulsiveness, In these chapters, we discuss interventions along the life lack of empathy), and cognitive deficits again in part due cycle since the ecological model suggests indeed that various to the school environment (low IQ, inattentiveness, poor types of interventions are needed from early childhood to school performance) are associated with violence and the completion of secondary school. delinquency in adulthood109. For young children, poor nutrition (stunting), exposure to toxic substance (such as RISK FACTORS lead or mercury), prenatal and perinatal complications, and head injuries can all lead to brain defects, The ecological model is a comprehensive framework that helps neurological dysfunctions, and learning disabilities, all in understanding and disentangling the complex factors leading of which may lead to violent conduct. For mothers, of violence and thereby identify some of its root causes. While substance abuse (alcohol and drugs) during pregnancy not all factors can be addressed within the confines of schools, increase risks for their children because they may also the model is useful to understand at a broad level some of lead to lack of neural/brain development and predispose the factors leading to violence, including VIAS. The model is children to aggression and antisocial behaviors110 . inspired by the public health literature and adopts the same Depending on context111 , other risk factors include 108 See WHO (2002) and Mercy et al. (1993). 109 Tremblay and Craig (1995). 110 See for example Murata et al. 2004) and Hu et al. (2006). 111 See Alda (2005) for a review. 112 Boys are more likely to be victims and perpetrators of physical violence (Schäfer and Korn, 2002; Sebastião et al., 2002). But girls are more often victims of dating violence, abuse, sexual assault and harassment, and exploitation. 113 See for example UNESCO and UNGEI (2014). 114 Alcaraz (2002). 47 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 gender112, sexual orientation113 , low socio-economic status, all, gang activity within schools and access to various and substance abuse. Age also matters114 as younger types of weapons are especially detrimental120 . students are more likely to be victims of violent behavior because of their inferior position in the balance of • At the community level, the presence of gangs is a powers between older and younger students. During major risk factor, as is the prevalence of crime and the early school years, some children engage in violent violence in the community. Availability of firearms, high behaviors such as kicking and hitting, spitting, and levels of unemployment or underemployment, high using pejorative nicknames for other children. Older population density, poverty121 , and norms conducive students are at greatest risk of initiation to violent crime. of violence and gender inequality also tend to be Still other risk factors include too much unstructured associated with higher risks of violence in schools. free time and association with delinquent peers115 . At the societal level, wars or dictatorships, norms condoning • At the family level, a wide range of factors may the use of violence to solve conflict, sometimes in the lead to victimization or perpetration of violence. In aftermath of political conflict, norms condoning gender particular, child maltreatment has been shown to have inequality such as patriarchal systems, and climates marked large effects on the propensity to be a perpetrator of by impunity and corruption all tend to be associated with violence later in life or be a victim of violence116 . It is higher risks of violence122. Violence as portayed in the media also associated with a higher risk of delinquency, and can also affect behaviors for both children and adults alike, more so than unemployment, educational attainment, as can weak police and legal systems and cultural norms gun ownership, or exposure to lead through paint or condoning violent behaviors, for example by husbands gasoline117. In addition, too soft or too harsh discipline at towards their wives and other dependents. home, corporal punishment, parental deviance, domestic violence, and the fact of witnessing violence at home, Important risk factors start as early as during pregnancy parental rejection, and poor supervision may all lead and the early years of a child’s life, with negative impacts to higher risks of violence later in life. Issues such as not only the child’s development, but also for future risks food insecurity may also be risk factors for violence. that the child may engage in violence, and lack the ability to be resilient to episodes of violence when victimized. • At the school level, risk factors include (among others) This suggests that a life cycle approach to preventing lack of resources, lack of organizational capacity, violence may be warranted. This is the approach used in and a school climate characterized by anti-social this study by identifying potential intervention in early behaviors. Lack of student bonding with their schools, childhood, as well as at the primary and secondary levels. use of corporal punishment in the school, and more Among different risk factors, both child aggression and generally a lack of positive role models and caring child maltreatment have been shown to have large effects adults also tend to result in higher risks of violence118 . in terms of future delinquency123 and aggression124 . Physical Drop-outs and student absenteeism and the presence aggression may appear as early as the first year of life of alcohol and drugs in the school are also negative and typically increases between 24 and 48 months of factors. Poor school governance and classroom age before decreasing as children learn to control their environment, including dilapidated infrastructure aggression125 . Longitudinal studies suggest that aggression and lack of safety making it easier for perpetrators at an early age is a strong predictor of aggression later in to enter schools are also detrimental119. But above life as well as other antisocial behavior including criminality. 115 Blum and Rinehart (1997) 116 See for example Mercy (2008) and Currie and Tekin (2012). 117 Chioda (2017). 118 See for example Watson (1995) and Rossman and Morley (1996).). 119 Fisher (2001). 120 Abramovay (2002). 121 Gottfredson (2001). 122 WHO (2002). 123 Currie and Tekin (2012). 124 Huesmann et al. (1984). 125 Tremblay (2004). 126 Chioda (2017). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 48 Figure 5.1: The Ecological Model Accumulation of context-based risk factors that evolve along a chronosystem INDIVIDUAL RELATIONSHIP SCHOOL COMMUNITY SOCIETAL Biology, personality Family abuse, parenting Low level of organization Presence of gangs and Inequality and poverty, (low self-esteem, greed, discipline, lack of a and poor school criminal organizations, impunity, higher impressionability), child positive role model, climate, poor student poor school system, tolerance towards maltreatment, victim or delinquent peers. connectedness with firearms, lack of violence in post-conflict witness of abuse as a child. schools, drop out. employment. countries. Source: Adapted by the authors from WHO . It may also lead to higher risks of unemployment and in life breeds success to prevent violence later in life127. mental health issues126 . Maltreatment of children and the children’s aggression tend to reinforce each other. PROTECTIVE FACTORS The various risk factors are context-based and change Each of the above risk factors has corresponding protective over time in the life of children. For an infant or a young factors that can be enhanced to both prevent episodes child, risk factors will mainly be at the individual and of violence and increase the resilience of children, family levels, while as the child grows and goes to school, families and schools to violence. Protective factors are he or she will face new forms of victimizations and risk defined as characteristics of the child, family, and wider factors associated with schools, peers and communities environment that reduce the negative effect of adversity will gain more salience. As such, adopting a life cycle on child outcome128 . At the individual and school levels, approach follows a developmental theory to violence protective factors include good student achievement, the prevention. If violence is a learned behavior, it can be ability of students to bond with the schools, strong work unlearnt and not learned in the first place and risk motivation, the ability in developed countries to reach factors contributing to it can be mitigated. Research higher education, support and supervision by teachers, shows that enhancing multiple protective factors early clear rules, and other positive features of the school 127 Tremblay and Craig (1995). 128 Masten and Reed (2002). 49 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 and class climate129. Research suggests that enhancing but also for fostering a sense of physical and social protective factors might be more effective to reduce order. Efforts to create a sense of school ownership VIAS than simply focusing on mitigating risk factors130 . among students as well as proper management of A more detailed discussion by level is provided below. access to various school areas (including minimizing opportunities for out-of-sight activities or blind • Individual and family-level protective factors include spots) can all foster a safe environment. Finally, caring relationships with parents and other adults maintaining high expectations also matters as both as well as positive communication and discipline, parents and schools that transmit high expectations all of which can help prevent violent behavior and to children and provide them with the necessary strengthen children’s resilience when violence occurs. support to achieve these goals have higher rates of Caring relationships also provide children with academic success, and lower risks of violence. positive role models and help create a strong sense of connectedness to both family and community while • Community and societal protective factors include promoting positive values131 . Students who develop participation in community networks. Students who higher self-esteem and self-efficacy are less likely to belong to groups such as neighborhood associations engage in violent behavior and acquaint themselves or religious and school organizations that foster the with delinquent peers. These students can also development of positive informal and formal ties adjust to change and recover from disruption and with other children, adults, and organizations are less stressful situations. Finally, children with outgoing likely to demonstrate aggression and/or violence133 . personalities – a trait which can to some extent be nurtured, may interact more easily with other children, The ecological framework is admittedly broad and at the which may again allow them to avoid engaging in operational level, more detailed framework could be used. violence as perpetrators or being victims of violence. At the World Bank, a forthcoming study is considering risk factors and interventions related to safety, teaching and • School-based protective factors include some of the learning, school-level relationships, and the institutional same caring relationships that work at the family level, environment. But while different framework can be used – but in this case with teachers and other school staff. whether at a strategic or operational level, identifying risk In addition, students who are given opportunities to and preventive factors has implications for programs and participate in the classroom and are recognized for policies. In addition, the accumulation of risk factors often their contributions, efforts, and progress in school, are explains why an individual behaves more violently or is more more likely to develop strong bonds of attachment and prone to be victimized than others134 . Therefore, instead commitment to school, family and community132. By of looking for the single best intervention that would be contrast, alienation from such processes in a student’s most effective to prevent violence, it often makes sense to development can result in the loss of bonds with combine interventions that would both mitigate the most the school, a potential precursor to violent behavior. salient risk factors and enhance relevant protective factors Indicators of school-level resources such as reasonable in a given context and for a specific age group. In other student/teacher ratios also can play a role. Teachers words, it often makes sense to work on at least two or three who have fewer students in the classroom can dedicate spheres of influence at once. Considering the different more time to each student and focus on those students spheres of influence of the ecological model, this means in that tend to interrupt the development of the lessons particular engaging families and communities apart from and the classroom. This helps for student engagement schools. Families play a crucial role in a child’s cognitive and and performance, thus reducing the risk of disorderly socio-emotional development, while communities convey behavior. Adequate physical infrastructure matters as norms that can condone or dismiss the use violence. well since schools that create a warm and welcoming These issues will be discussed in more details in the next environment not only help for the student experience, two chapters. 129 Lösel and Farrington (2012). 130 Resnick (2000). 129 Blum et al. (2000). 130 Hawkins et al. (1991). 131 https://www.blueprintsprograms.org/publications/ 132 Olds et al. (1998); Farrington and Welsh (2007). 133 Catalano et al. (1999) and Violence Prevention Institute (2001). 134 World Health Organization (2002). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 50 CHAPTER 6 including community service. Another review136 used different categories: (i) instructing students; (ii) behavior PROMISING SCHOOL- modification and teaching think strategies; (iii) mentoring, tutoring, and work-study experiences; (iv) counseling and BASED INTERVENTIONS social work; (v) recreational, community service and leisure activities. And school environment interventions, including As mentioned in the previous chapter, the accumulation (vi) managing schools and disciplines; (vii) establishing of risk factors often contributes to either perpetrating norms and behaviors; (viii) managing classes; and (ix) or being a victim of violence. Rather than relying on regrouping students. Still other reviews have proposed a single intervention to prevent violence, it is often different typologies. Some typologies may work in some better to combine interventions to tackle risk factors context, others in other contexts. In this chapter, we rely and enhance protective factors. This chapter focuses on on a life cycle approach to present successful interventions. school-based interventions within a life cycle approach. School-based violence prevention interventions can Risk factors evolve over time, but research shows that be effective, but their effectiveness vary by age group enhancing multiple protective factors early in life helps to and the type of violence and bullying being observed. prevent violence later in life. However, it is never too late The chapter showcases examples of programs that have to prevent violence, and interventions later in life have a been rigorously evaluated and have proven effective in role to play. Table 6.1 provides the typology that we rely preventing different types of VIAS. Interventions are on to organize this chapter and the next, together with a suggested according to the age groups for which they few examples of effective programs by level of education. are likely to have the most impact, even though several interventions can be effective for different age groups. EARLY CHILDHOOD: PREVENTING CHILD While many of the interventions have been evaluated in MALTREATMENT AND ENHANCING developed countries, an emerging body of evidence is PARENTING SKILLS now also available for developing countries. It is important to note that when planning interventions, considering As mentioned in chapter 5, risk factors start during “how to” is important beyond considering “what to” do. pregnancy and the early years of a child’s life. Therefore, Although programs to end violence in school could be interventions to prevent VIAS should begin with prenatal considered as part of a package of essential school health care and selected other ECD programs137. In Rio, child care interventions, discussing components of such a package programs have been shown to improve child assessments is beyond the scope of this study. Also, in addition to and lessen behavioral problems138 . The Colorado Blueprints interventions, broader systems at the level of a country for Healthy Youth Development139 identifies three are needed especially to reduce gender-based violence effective programs targeting 0-4 years old (infants and and sexual exploitation, abuse, and harassment. preschool children) and 16 promising programs. While not reviewing all programs here, a few can be highlighted. LIFECYCLE APPROACH A first well known intervention focusing on healthy There is no unique or best way to categorize programs pregnancy and parenthood readiness is the Nurse Family that seek to prevent VIAS. One review135 classified Partnership in the United States. This home visitation interventions in nine categories: (i) enhancement of program accompanies women during their first pregnancy academic skills; (ii) after-school programs; (iii) mentoring/ to prevent substance use (alcohol, tobacco, or drugs), monitoring programs; (iv) social skills training for promote a healthy diet, and equip them with positive students; (v) skills training for teachers; (vi) school-wide parenting skills. It also focuses on raising knowledge interventions; (vii) violence reduction; (viii) counselling and of infants’ needs, including awareness of the dangers mental health interventions; and (ix) other interventions of exposure to toxic substance. The program helps in 135 Valdebenito et al. (2019). 136 Gottfredson et al. (2004). 137 Denboba et al. (2014). 138 Carneiro and Evans (2013). 139 https://www.blueprintsprograms.org/publications/ 51 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Table 6.1: Typology and examples of effective programs by level of education Early Childhood and Primary Secondary Examples of Preschool (0-5) (6-11) (12-18) Effective Programs Integrated Approaches Home Visitations Nurse Family Partnership Whole School Approach Good School Toolkit Anti-Bullying Olweus anti-bullying programs, KiVa bullying prevention Curriculum Enhancement Socio-Emotional Learning Promoting Alternative Thinking Strategies (PATHS) Gender Equality Fourth R (Preventing Dating Violence); Programa H Conflict Resolution Aluas con Paz Mindfulness Quiet Time; Mindful Schools Cognitive Behavioral Training BAM; EmpaTeach Mentoring BAM After-school Programs Right to Play Empowering Key Stakeholders Teachers' Skills Enhancement Irie Classroom Toolbox; Good Behavior Game; (positive discipline; classroom Good School Toolkit management) Parenting Functional Family Therapy; Multisystemic Therapy Community Norms Bell Bajao; Good School Toolkit; SASA! Source: Authors. preventing child maltreatment and increasing bonding ratio), and had weekly home visitation and parenting with infants and positive discipline. It also helps to increase program to involve parents in the socio-emotional the importance given by mothers to early child care. development of their children141 . Children who participated Evaluations show significant positive outcomes in both in the program were followed up to age 27. Positive the short and long term. In the short term, substantial impacts included higher educational performance, reductions in child abuse and neglect were observed. In including higher rates of high school completion, lower addition, in the long term, a 15-year follow-up assessment involvement in delinquency (fewer lifetime arrests), showed reduction in serious antisocial behaviors as and improved economic status when adults142. measured through the number of arrests and convictions in comparison to control groups. Effects were even greater The Child-Parent Center Education Program in for children from disadvantaged families, suggesting Chicago targeted inner city children aged 3 to 9. the benefits from well targeted interventions140 . Participating children showed higher educational attainment, income, socioeconomic status, as well as The High/Scope Perry Preschool program is another lower rates of justice-system involvement and substance well-known program that has been rigorously evaluated abuse later in life. The program emphasized basic skills and has strong results. Disadvantaged children attended in language arts and math through structured and the program daily for 2.5 hours per day for two years, diverse learning experiences that included whole-class from 3 to 5 years of age and a total of 30 weeks a instruction, small-group and individualized activities, and year. The curriculum focuses on stimulating cognitive frequent field trips. All teachers were certified and had development and enhancing self-control skills, was bachelor’s degrees. Classes were small and staffed by organized in small groups dynamic (5:1 teacher-children aides. In addition to the head teacher in each site, the 140 Olds et al. (1998); Farrington and Welsh (2007). 141 Tremblay and Craig (1995); Heckman et al. (2010). 142 Schweinhart et al. (1993). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 52 Figure 6.1 Social-emotional learning Self-awareness Responsible decision Self-management making Relationship skills Social awareness Source: Based on Weissberg et al. (2015). parent resource teacher and outreach representative Generation PMTO, a family training program focused directed multi-faceted and intensive services in the on family management skills. To ensure quality parent resource room. The scope of services helped implementation of evidence-informed early childhood ensure high participation. Heavy outreach by staff interventions, performance checklists (lists of tasks or also led to participation by families most in need143 . steps required to complete a practice competently) such as those developed by the Early Childhood Technical Parent Corps is an evidence-based intervention that Assistance Center at the University of North Carolina enhances pre-K programs in schools and early education - Chapel Hill can be a valuable tool. A meta-analysis of centers, serving primarily children of color from low- performance checklists for ECD interventions in the income communities. It helps key adults in children’s lives United States showed that the 26 checklists developed to — parents and teachers — to create safe, nurturing, and implement best practice ECD programs were based on predictable environments at home and in the classroom solid practice-outcome relationships. That is, they used and improve relationships and communication between strong evidence and findings from research syntheses and parents and teachers. ParentCorps includes three main empirical studies to inform the selection or development components: professional learning for leaders, teachers, of checklist indicators.145 This suggests that performance mental health professionals, and parent support staff; checklists might be relevant and reliable tools, in parenting program for families of pre-k students, particular in efforts to adapt or scale up programs. and social-emotional learning classroom curriculum. Among various outcomes, children showed less child Finally, healthy nutrition in early childhood is also critical. internalizing and externalizing problems at two-year Poor nutrition and especially stunting can lead to long-term follow-up (age 8) than those in the control group144 . damage to the brain, which, in turn, can impact a child’s emotional and psychological responses to stress, learning Other programs shown to be effective for children’s disabilities and other medical complications. This may healthy development include Project STAR and affect educational performance and achievement, as well 143 Reynolds et al. (2011). 144 Brotman et al. (2013); Dawson-McClure et al. (2015); Brotman et al., (2016). 145 Dunst (2017). 53 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 as productivity in adulthood. Later in the child’s life, once primary education. The Promoting Alternative Thinking healthy nutritious habits are acquired, they can last beyond Strategies (PATHS) in Jamaica focused on enhancing the school years and into adulthood. Insights are available socio-emotional skills, in particular self-control and on new interventions that focus on nutrition, mental emotions regulations through a multi-component health, and mindfulness, as a tool to curve violence146 , programs in primary schools (5-11 years old students), especially in primary schools to which we now turn. with strong results in enhancing protective factors such as better educational and employment outcomes150 . POSITIVE EMOTIONS MANAGEMENT AND PEACEFUL CONFLICT RESOLUTION Mindfulness is a relatively simple and low intensity technique that has shown positive results in terms of stress Socio-emotional learning (SEL) programs are a common reduction, self-control but also reduction of suspension and and effective approach to prevent VIAS. As show in grades improvement. In the United States, mindfulness has Figure 6.1, SEL and socio-emotional skills can be broadly been introduced in elementary and middle schools through categorized in five groups of skills related to self-awareness short sessions of breathing exercise and learning to be in (understanding one’s own emotions, personal goals, and the present moment. Evaluations and systematic reviews values, as well as one’s strengths and limitations), self- consistently found positive outcomes151 . In San Francisco, management (regulating one’s emotions and behaviors, the program Quiet Time that involves two 15 minutes for example by delaying gratification, managing stress, sessions of breathing meditation per day (one in the controlling impulses, and showing grit), social awareness morning and one in the afternoon) has led to a significant (understanding, empathizing, and feeling compassion drop in anti-social behavior in school and outside school. others, including those with different backgrounds or This includes a reduction of 45 percent of suspensions and cultures), relationship skills (establishing and maintaining grades improvement152. A similar intervention implemented healthy and rewarding relationships), and responsible in Oakland called the Mindful Schools had similar decision making (make constructive choices about personal positive results on stress reduction and self-control153 . behavior and social interactions across diverse settings)147. SEL seeks to empower students to manage their emotions, Some programs focus on peaceful conflict resolution. achieve positive goals, empathize with others, sustain The Aulas en Paz program seeks to reduce and prevent positive relationships, and make responsible decisions148 . aggression and promote peaceful coexistence in Colombian A related approach to SEL emphasizes conflict resolution schools. The program consists of a curriculum to develop skills for students. Programs focusing on the skills can competencies in the classroom, extracurricular activities help students reduce instances of interpersonal conflicts, in groups of children identified initially as aggressive with thereby preventing or reducing violence and bullying. children identified to having more prosocial skills, and workshops, visits, and regular phone calls to parents. A review of four meta-analyses of school-based SEL More than 80,000 children have benefited from the programs suggests many positive outcomes including lower program in elementary schools. An evaluation suggests levels of conduct problems and emotional distress and reductions in levels of aggression and indiscipline and higher academic performance for children participating an increase in prosocial behavior among beneficiaries. from preschool through secondary education. Even The success of the intervention, particularly in violent though immediate results were stronger, positive contexts, appears to be predicated on the mix of universal outcomes persisted in the long term in a variety of activities for the general school population and targeted contexts149. While SEL programs have been successful activities for those who need more targeted attention154 . in enhancing prosocial behaviors across age groups, some interventions had particularly strong results in 146 Chioda (2017). 147 Weissberg et al. (2015); Oberle et al. (2016). 148 Weissberg and Cascarino (2013). 149 Mahoney et al. (2018). 150 Chioda (2017). 151 Napoli et al. (2005); Weare (2013); McKeering and Hwang (2019). 152 Hözel et al. (2011). 153 Chioda (2017). 154 Jimenez et al. (2010). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 54 Steps can be taken in primary school to prevent gender- sub groups of children by gender, ethnicity, developmental based violence in adolescence. Gender-based violence needs, disability and other characteristics.155 A meta- results in part from cultural norms, traditional gender analysis on school-based violence prevention programs in roles in the family and society, and power imbalances the United States also suggests that programs focused on within the family, community, and society. School at-risk and older children had stronger effects in reducing violence prevention programs must incorporate a gender violence156 . This is in line with research on conflict resolution lens. For example, if a school is designed properly, the education mentioned earlier that tends to be more risk of gender-based violence can be mitigated. effective for older than younger children157. After-school programs also tend to be more effective for adolescents158 In addition, the school is a place where cultural norms can probably due to positive peer associations. However, be challenged and reshaped to support gender equality mixing students is not always effective probably because and prevent gender-based violence. Even when teaching of the same yet reverse reason (working with peers can traditional academic subjects, teachers can strive to be counterproductive if this is associated with increased promote a culture of non-violence, as well as equality victimization and even a potential increase in bullying)159. and respect for different genders and sexual orientations Remedial education and life skills and training programs for both within the school and the wider community. very high-risk adolescents also show positive outcomes160 . SECONDARY EDUCATION: ANTI-BULLYING The Campbell meta-analysis of 44 evaluations of anti- INTERVENTIONS AND GENDER EQUALITY bullying programs reviewed studies with evidence of impact PROGRAMS conducted from 1983 to 2009161 . It shows that on average, the programs achieved to reduce bullying by 20-23 percent Some interventions tend to be more effective with and victimization by 17-20 percent. Program components older children, especially to prevent bullying defined as that proved more effective to reduce bullying included: repeated aggression (physical, verbal or psychological) parent training/meetings, improved playground supervision, among peers who have an imbalance of power (one child disciplinary methods, classroom management, teacher being more powerful than the other) over a prolonged training, classroom rules, whole school anti bullying policy, period of time. In Europe, anti-bullying programs have school conferences, information for parents, and cooperative been most successful with children aged 11 or older, even group work. The most effective components were parent if more research is needed on differentiated impacts for trainings/meetings and disciplinary methods, as well as 155 Farrington and Ttofi (2009). 156 Park-Higgerson et al. (2008). 157 Garrard and Lipsey (2007). 158 Gottfredson et al. (2004). 159 Farrington and Ttofi (2009). 160 Guerra et al. (2013). 161 Farrington and Ttofi (2009). 55 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 intensity for children and parent training/meetings. For approaches such as cognitive behavioral training (CBT) victimization, videos and disciplinary methods were the have proven effective to lessen impulsiveness and most effective components, followed by parent training/ enhance empathy. CBT challenges automatic responses meetings and cooperative group work. Among the different and questions triggers that often leads to unwelcomed components, enhancing playground supervision, with consequences. The Colorado Blueprints identify more teacher presence, identifying hot spots and hot six model interventions that involve CBT: LifeSkills times for bullying, and re organizing the built environment Training, Blues Program, Brief Alcohol Screening and and natural surveillance (supervision) accordingly were Intervention for College Students, New Beginnings (for also promising and relatively low-cost interventions. Children of Divorce), Project towards no Drug Abuse, and Promoting Alternative Thinking Strategies. In Comprehensive programs inspired by the Olweus model addition, the Colorado Blueprints identify 12 promising - which aims to address bullying and improve pupil CBT programs. The focus of the discussion below is on relationships from elementary school to high schools, tend impacts on delinquency and physical violence. CBT may to work well. The program has been implemented in a also mitigate the effects of child sexual abuse, including dozen countries. Students respond to a short anonymous depression, post-traumatic stress and anxiety, but a meta- survey to assess the nature and prevalence of bullying in analysis suggest that those impacts are generally modest. the school. A conference day is then convened for school staff to discuss the results of the survey with support The LifeSkills Training (LST) program has been rigorously from consultants. A Bullying Prevention Coordinating evaluated multiple times with consistent, strong evidence Committee with representation from all key stakeholders of its effectiveness to prevent violence and substance is created and an action plan is designed. Actions include abuse for high school students. LST uses multiple methods adult supervision of school areas where bullying tends to (instruction, demonstration, feedback, reinforcement occur. Regular classroom discussions and activities are and practice) to teach high school students personal held to reinforce rules and anti-bullying values and norms. self-management skills, social skills, and information The program also encourages parental involvement. and resistance skills related to drug use. A total of 30 Finally, some steps target students who bully and those sessions are taught over three years with a decreasing who are bullied, as well as their parents. Evaluation number of sessions over time (15, 10, and 5 sessions per suggests that the program may reduce bullying by half162. year). Additional violence prevention lessons also are available each year. Results from a large-scale randomized For such programs to work, attention must be paid to study in New York City showed a reduction of 32 the comprehensiveness of the program (number of percent in delinquency and 26 percent in the likelihood components), as well as its duration (number of days) and of fighting after the first year of the program164 . intensity (number of hours) for teachers and children influence effectiveness towards bullying reduction. Becoming a Man (BAM) targets at-risk adolescent boys Intensive and long-lasting program are needed to change from very disadvantaged neighborhoods in Chicago. It behaviors and norms in and around schools, which matters focuses on emotions management, interpersonal problem for prioritization when designing school programs. As solving, goal setting, empathy, and personal integrity discussed in the next chapter on stakeholder engagement, using CBT and role modeling in and after school. Youth engaging with families and communities is also key for meet on a voluntary basis every week with a mentor for a success. Parents sessions were particularly effective for check-in during which they can share personal concerns. both reducing bullying and preventing victimization163 . The group listens in and situations are reviewed in groups to analyze responses and alternatives. Those sessions are For at-risk adolescents that display serious anti-social followed up with sport sessions with trained coaches. behaviors, are involved in delinquency and/or are at Sports activities help reinforce SEL and positive conflict the verge of dropping out of secondary school, other resolutions with specialized guidance. BAM lasts 27 162 Hazelden Foundation (2007). 163 Farrington and Ttofi (2009). 164 Botvin et al. (2006). 165 Heller et al. (2013). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 56 hours with weekly group sessions over the school year. students’ motivation, engagement to school, and more Evaluations show positive short-term effects not only generally prosocial and less aggressive behaviors in life. in reducing antisocial behaviors, but also in improving Positive mentoring also had multiple positive effects in academic achievements. Specifically, BAM helped promoting prosocial behaviors as they provide positive role reduce violent crime arrest by 44 percent and arrests models and positive connectedness with a caring adult. related to vandalism and weapons crimes by 36 percent. In addition, participants were more likely to graduate Specific interventions focus on preventing dating from high school compared to non-participants 165 . violence. Adolescents are particularly vulnerable to reproducing patterns of violence witnessed at home or in the community in terms of gender-based violence. Mentoring, teachers’ skills enhancement, counselling and Effective programs aim to empower adolescents to build mental health services, and academic support have also and maintain healthy relationships. They usually include been found effective to reduce exclusion, and thereby avoid knowledge on how to detect abuse and develop skills to its multiple negative consequences on both adolescents’ ensure respect in romantic relationships. Interventions learning achievements and violence perpetration and such as Fourth R in Canada aim to empower adolescents victimization. Disciplinary exclusion has been linked to to build and maintain healthy relationships. They usually antisocial behaviors and delinquency166 as well as poor include knowledge on how to detect abuse and develop learning outcomes, including dropout167 with long-term skills to ensure respect in romantic relationships. An negative outcomes in terms of employment and training evaluation suggests that students, particularly males, who opportunities168 . Meta-analysis found that skills training for participated in the program at aged 14-15 were significantly teachers and mentoring programs were the most effective less likely to perpetrate acts of violence towards their (in the short term) to reduce exclusion169. Other effective partners two years following program participation at age interventions included enhancement of academic skills 16-17, compared to peers who did not participate170 . Other and counselling/mental health services. These findings programs focusing on sexual education have also shown are in line with other research that emphasized the positive results in reducing coercive sex between students importance of teacher-student relationships to enhance and fostering higher resilience to pressure for having sex. 166 Hemphill and Hargreaves (2010). 167 Arcia (2006); Michael (2011); Noltemeyer and Ward (2015). 168 Massey (2011). 169 Valdebenito et al. (2019). 170 Wolfe et al. (2009). 171 Rijsdijk et al. (2011); UN-Women and UNESCO (2016). 57 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 This included the World Starts with Me in Uganda that and trained counselors or focal points in schools. targeted both male and female students aged 12-19171 . Raising awareness on the need to report is not an easy Finally, after-school programs usually combine recreational task, however. Data from surveys measuring violence activities (sports or arts) and academic support to against children suggest lack of reporting of episodes promote adolescents’ positive use of free time, positive of violence, both in school and elsewhere. In Kenya and peer association, and mentoring while strengthening Uganda for example, between 40 and 60 percent of socio-emotional learning, gender equality, and life boys and girls who were victims of VIAS told someone skills. When reviewing after-school programs and their about the incident, but less than 10 percent of victims respective effectiveness, attention should be paid to the receive services174 . Part of the issue is a lack of awareness diversity of programs in terms of form, structure and about the services that are needed or available. Another specific goals, and the need to focus on quality.172 Some issue is the high level of normalization of violence after school programs provide intensive, small-group against children. There is, therefore, also a need to teach instruction or individual tutoring programs, while others children, including young boys and girls, to identify only provide unstructured homework time, which is abuse. But in addition, lack of reporting is also due to unlikely to lead to significant results. Provided they follow lack of trust that appropriate actions will be taken and evidence-based practices, after-school programs can perceptions that no retaliation again perpetrators will lead to significant reductions in conduct problems and happen. In some settings, there may also be incentives drug use, as well as gains in achievement (as measured for principals and teachers not to report violence in their through test scores), grades, and school attendance. school, for example to preserve the school’s reputation. Researchers summarize these practices as SAFE, which stand for (S)tep-by-step training approach, (A)ctive When reporting incidents of violence, adequate referral forms of learning by having youth practice new skills, pathways are needed. Referral pathways are protocols (F)ocused time and attention on skill development, (E) for a series of actions or steps to take after identifying a xplicit in defining the skills to be promoted. After-school student who has been the victim of violence and bullying. programs tend to be more effective for adolescents, Their primary objective is to ensure that survivors/ probably because they enable positive peer association173 . victims of violence and bullying receive prompt and coordinated responses from service providers while REFERRAL PATHWAYS: WHAT TO DO WHEN also ensuring the safety and privacy of the victim. CASES OF VIOLENCE ARE IDENTIFIED The services provided must be comprehensive and may include health services, psychosocial support, The interventions mentioned so far focus on prevention. protective care (police services), and legal services if When violence or bullying occurs in schools, the needed, as well as linkages to other community service students who have been victimized need support. providers. During this process, school actors must Child-friendly, safe, and reliable reporting mechanisms ensure that existing policies and agreed procedures are and referral pathways must be in place to allow an followed, including for prosecution of perpetrators. ethical and timely response when a child experiences violence. This is crucial to prevent further trauma and Ensuring the confidentiality of the survivor/victim is critical revictimization, provide children with the right support as is maintaining high ethical standards during the process and services, and prevent perpetrators from continuing of referral. To that end, school may want to identify and doing harm. Different modalities exist to enable safe train focal points for both genders within the schools so reporting in school, including child-friendly helplines, that student victims/survivors can reach out directly to chat rooms and online reporting, anonymous boxes, them in order to maximize the effectiveness of the referral 172 Durlak et al. (2010); Hirsch et al. (2010). 173 Hill et al. (2011); Chioda (2017). 174 Catholic Relief Services (2019). 175 UN-Women and UNESCO (2016). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 58 pathway system. Many countries have adopted referral Sexual harassment may occur when it interferes with pathways, including Guatemala and Peru, as well as Kenya work, is made a condition of employment or creates an and Uganda. In Kenya, a free National Child helpline intimidating, hostile or offensive work environment. It may was set up with trained volunteer counselors and a clear also occur in public and private spaces and in schools. referral pathway with a list of service providers specialized in child services. More research is needed on how The share of countries without laws on sexual harassment effective referral pathways are in practice, but guidance is decreasing. Between 2013 and 2017, it dropped by three on how to choose reporting mechanisms is available175 . percentage points from 16.3 percent to 13.5 percent thanks to legal reforms in Cameroon, Chad, Egypt, and Guinea. Similarly, when considering other areas, some BEYOND SPECIFIC PROGRAMS: gains were observed as well. At the same time, one in THE IMPORTANCE OF POLICIES AND LAWS five countries does not have appropriate laws against sexual harassment in employment. The proportion is The discussion so far is based on experimental or quasi- much higher at six in ten countries for sexual harassment experimental evaluations and focuses on the empirical in education and at four in five countries for sexual evidence on programs that appear to work to reduce harassment in public spaces. Criminal penalties for sexual VIAS. Beyond specific programs, it is also important to harassment, such as fines or imprisonment are in place in mention the importance of broader policies, including laws. only two thirds of countries and less than half for sexual When policies are implemented well, and national laws harassment in employment. There is heterogeneity have adequate enforcement mechanisms, they can have between regions in the share of countries with adequate a major impact towards reducing violence at scale, which laws, with weaker laws for some (but not all) categories may in turn reduce the need for specific programmatic in the Middle East and North Africa as well as sub- interventions. The referral pathways mentioned above is Saharan Africa. In nearly nine out of ten countries, laws one example of broader policies, as are codes of conduct covering sexual harassment do not specifically apply to in schools adopted by school systems. Laws to prevent the different domains of employment, education and the use of corporal punishment policies in school are public places combined, suggesting that most countries another example of national policies that can make continue to have at least some gaps in their laws. a difference at scale. Yet while many countries have adopted laws to prevent corporal punishment, including Despite more countries adopting laws against sexual in school, the practice often remains used by teachers harassment, including in education, the number of due to weak enforcement. This does not mean that laws women not protected under the law is decreasing should not adopted, but rather that actual practices only slowly, in part due to population growth. In 2017, should be monitored and sanctioned when needed. 287 million adult women were not legally protected from sexual harassment, versus 320 million in 2013. To illustrate how laws related to violence in schools remain The share of women not protected was reduced from inadequate in many countries, consider an analysis recently 12.5 percent to 10.6 percent. But when considering conducted using data from the Women, Business, and specific domains, the statistics are worse. Estimates of the Law program at the World Bank176 . The analysis the number of women lacking legal protection against documents global trends in legal protection for women sexual harassment in employment, education, and public and girls against domestic violence and sexual harassment. places are at 359 million globally, 1.5 billion, and 2.2 In particular, Women, Business and the Law collects billion, respectively. These estimates are all very high. data on legislation on sexual harassment, defined as any unwelcome sexual advance, request for sexual favor, verbal or physical conduct or gesture of a sexual nature, or any other behavior of a sexual nature that might reasonably be expected or be perceived to cause offence or humiliation. 176 Tavares and Wodon (2018). 59 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 been discussed as part of the previous chapter since CHAPTER 7 they are also good examples of specific interventions. The message we wish to give by providing guidance in ENGAGING ALL two separate chapters – one on interventions, and the other on the need to engage with stakeholders, is that STAKEHOLDERS while specific interventions are identified as promising, all interventions should be implemented with a view, when Multiple stakeholders must be engaged to ensure that feasible, to engaging relevant stakeholders. Interventions schools are safe. This includes stakeholders within the without broad engagement with stakeholders are not school – not only students, but also teachers and principal likely to succeed, and similarly engagement without – as well as stakeholders in the community, including specific (evidence-based) interventions is also unlikely to parents and caregivers. There are examples of successful succeed. Both specific interventions and engagement with programs adopting so-called whole school approaches stakeholders are needed. As mentioned in the previous to violence prevention, as noted by WHO. One of the chapter the “how to” matters apart from the “what to” objectives of engaging with the school community at focus on to end violence in schools. Beyond specific large is to change mindsets on VIAS (and more generally stakeholders, governance and laws, reporting mechanisms, violence against children), while also challenging where and systems, as well as codes of conduct, all matter. This is needed patterns of gender inequality that sustain gender- in particular the case for laws pertaining to gender-based based violence. This chapter considers ways to strengthen violence and sexual exploitation, abuse, and harassment. teacher training and engage parents/caregivers so that they fully participate in programs. Examples of programs WHOLE SCHOOL APPROACH that have managed to change community norms towards the use of violence are also provided. While these A whole school approach is a comprehensive approach programs are discussed in this chapter because on their that seeks to change norms at the school level to emphasis on engaging stakeholders, they could also have promote respect, inclusion, good behavior, and learning 176 Tavares and Wodon (2018). 177 CEAPA (2012). 178 Bradshaw et al. (2012); Cornell et al. (2012); Lewis et al. (2013); Ward and Gersten (2013); Sprague et al. (2016); Un-Women and UNESCO (2016); WHO (2019). 179 Frankenberg et al. (2010); Payet and Franchi (2008). 180 Durrant (1999). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 60 and safety. This is done by engaging with the entire campaigns helped decreased public support to corporal school community, including pupils/students, principals, punishment from 54 percent to 11 percent in 25 years180 . teachers, school administrative staff, counsellors, parents, and community members and leaders. Potential In primary schools, a well-known program using a whole- components of a whole school approaches have been school approach is the Good School Toolkit (GST) in defined by various organizations in slightly different Uganda, a country with substantial structural issues related ways, but with similar intent. Beyond the prevention to poverty, large class sizes, poor physical infrastructure, of VIAS, the Center for Disease Control considers 10 and a lack of resources for teaching. The program was components in its Whole School, Whole Community, developed by the NGO Raising Voices in consultation Whole Child (WSCC) approach: Physical education and with teachers and children in 600 schools. It works physical activity; Nutrition environment and services; through six sequential steps to provide behavioral change Health education; Social and emotional school climate; techniques to teachers, school staff and students. It Physical environment.; Health services; Counseling, focuses on goal setting, positive discipline, empathy, and psychological and social services; Employee wellness; reflection and practice of new behavioral skills. Teachers Community involvement; and Family engagement. and school staff are trained and supported throughout The United Nations Girls Initiative suggests that core implementation. An evaluation suggests that after 18 elements of a whole school approach towards ending months of implementation, GST reduced the risk of gender-based violence should include: Effective school physical violence by teachers and school staff against leadership and community engagement; Establishing students by 42 percent; halved the number of teachers and implementing a code of conduct; Capacity building who reported using physical violence against students; of teachers and educational staff; Empowering children and improved students’ connectedness and sense of on child rights, participation, and gender equality; safety and belonging with their school. GST also increased Improving reporting, monitoring, and accountability; teachers’ satisfaction in their role at school and increasing Addressing incidents; Strengthening physical learning students’ wellbeing and sense of safety at school181 . environments; and Engaging parents. These two lists are not exhaustive: other ways of conceptualizing ENHANCING TEACHERS’ SKILLS ON POSITIVE the whole school approach have been proposed177. DISCIPLINE AND CLASSROOM MANAGEMENT A whole school approach uses multiple strategies to Supporting teachers to enhance their skills in positive develop a common vision and shared values and rules discipline and classroom management is one of the most for the school, and works through the curriculum, effective approaches to preventing bullying and violence teacher training, parental engagement, and student in and around school. Among the different types of learning towards a safe and inclusive school climate school violence prevention interventions reviewed in a and respectful school values178 . A key characteristic of meta-analysis182, those focusing on providing teachers a whole school approach is thus collaboration among with skills to improve their relationship with students main school stakeholders. Teachers and parents play and manage students’ behaviors had the strongest and an especially crucial role. One of the objectives is to most reliable results in terms of lessening disruptive and change mindsets on corporal punishment and violence aggressive behaviors in the classroom and enhancing against children as a way to educate children179 as well as prosocial behaviors later in life. By contrast, punitive gender inequality that sustains root causes for violence. interactions tend to feed a vicious circle of violence, Breaking these habits and cognitive frames is not easy, delinquency, and further exclusion. These results are in but it can be done possible. In Sweden for instance, line with other systematic reviews and meta-analyses183 . a general ban on all forms of corporal punishment The types of training reviewed in the meta-analyses against children with corresponding public education included training focusing on facilitating mutual respect 177 CEAPA (2012). 178 Bradshaw et al. (2012); Cornell et al. (2012); Lewis et al. (2013); Ward and Gersten (2013); Sprague et al. (2016); Un-Women and UNESCO (2016); WHO (2019). 179 Frankenberg et al. (2010); Payet and Franchi (2008). 180 Durrant (1999). 181 Devries et al. (2015); Naker (2018); WHO (2019). 182 Valdebenito et al. (2019). 183 Oliver et al. (2011); Obsuth et al (2016); Allen et al. (2007); Cornelius-White (2007); Roorda et al. (2011). 61 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 between teachers and students184 , training focusing management techniques, and develop behavior planning. on class management and establishing clear classroom Evaluation showed reduction in teachers’ use of violence rules185 , and training providing teachers with strategies against teachers and overall improvement in children’s for working in alliance with parents to promote students’ prosocial behavior189. Another promising program for engagement in school activities186 . SEL training should be primary teachers is EmpaTeach developed by the included in teacher training given its positive effects on International Rescue Committee. This is a low- intensity students’ prosocial behaviors and learning achievements187. 10-week intervention during which teachers are taught strategies to maintain discipline in their classrooms Several programs have been effective at empowering in nonviolent ways. EmpaTeach focuses on teachers teachers at different levels of education to use positive themselves using cognitive-behavioral therapy and discipline and improve trust and mutual respect in behavioral science techniques to help them identify their interaction with children. The Global Initiative their triggers, change destructive thought patterns, to End All Corporal Punishment of Children and the and plan for positive reactions. Pilot findings showed WHO handbook on school violence prevention provide a reduction in physical and emotional VIAS (as self- useful resources on positive discipline for teachers reported by students), higher student connectedness, less and schools188 . When teachers and the entire school depression, and more engagement in school. The use of community understand that respect and trust are the empathy building module also succeeded in changing key pillars for child’s healthy development and that teachers’ attitudes towards corporal punishment.190 corporal punishment is not only counterproductive but negatively impacts a child’s learning ability, the whole ENGAGING WITH PARENTS AND CAREGIVERS school culture and climate may be transformed. Since many risk factors associated with aggression and At the preschool level, the IRIE Classroom Toolbox violence are the individual and family levels, families has empowered early childhood teachers to create need to be part of school programs as they keep playing emotionally supportive classroom environments, teach an important role in children and teenagers’ life and preschool children socio-emotional skills, use classroom development. In fact, engaging with parents of adolescents 184 Okonofua (2016). 185 Hawkins et al. (1988). 186 Ialongo et al. (2001). 187 Mahoney et al. (2018); Schonert-Reich et al. (2017). 188 WHO (2019). 189 Baker-Henningham et al. (2017). 190 International Rescue Committee (2017). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 62 that display behavioral problems can yield significant Parenting programs need to follow evidence-based results in a relatively short period of time. For instance, the practices and focus on positive discipline, positive Functional Family Therapy program in the United States communication, and increased bonding among family works with the entire family in 12 sessions over 3 months members. As with teachers, providing alternative tools and to increase bonding, problem solving and communication. skills to caregivers in dealing with their children can help Results suggest a large reduction in recidivism for youth to break the cycle of violence. Child maltreatment is a participating in the program (79 percent reduction after 3 major risk factor for perpetrating or experiencing violence years191). Another well-known program is the Multisystemic later in life, leading to multiple negative outcomes193 . Therapy which has also succeeded in reducing behavioral Even mild corporal punishment can be detrimental problems. The program combines CBT, parenting programs to children’s healthy development and wellbeing, with to promote affection and communication, after school negative multiplier effects in terms of externalizing and activities, and increased parental involvement in the internalizing behavior problems, impaired cognitive ability, education of children and youth (12-18 years of age)192. low self-esteem, and the normalization of violence against children.194 Communication within the family is also 191 Gordon et al. (1995). 192 Weiss et al. (2013); Timmons-Mitchell et al. (2006); Dekovic et al. (2012); Asscher et al. (2014). 193 Zingraff et al (1994); Hagborg et al. (2018). 194 Gershoff (2002); Gershoff and Grogan-Kaylor (2016); Font and Cage (2018). 63 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Figure 7.1: Engaging with Parents and Caregivers Clear & wide recruitment process Focusing on trust with Good communication and parents liaison Removing barriers to Incentives for recruitment participation and participation Active and creative outreach Source: Adapted from Axford et al. (2012). important to identify potential experiences of violence or among the many parenting methodologies, even though bullying that may be ignored if there is no communication quality matters, but to keep them engaged long enough or trust in the family. However, not all programs are equal, to produce sustained behavioral change. A few lessons with different types, structures and intensity affecting emerge from research and practice (Figure 7.1)199. A outcomes.195 While some parenting interventions have thorough recruitment process, good communication step-by-step protocols and resources and tools with and liaison with stakeholders, incentives for recruitment sessions in the classrooms and sharing of practical how- and retention (such as in-kind incentives for parents200 to information with parents and caregivers, others may but also rewards for recruiters201), active and creative simply consist of newsletters or unstructured engagement outreach work providing flexibility for practitioners with uncertain results. Several resources on promising to innovate, investment in building relationships with examples of parenting programs are available including the parents, making programs easily accessible, and having WHO School Violence Handbook196 , the UN Women’s realistic expectations can all help making parenting guidance on school-related gender-based violence197, programs work. Other studies confirm these best the INSPIRE Handbook and associated resources198 , practices and show that building relationships with and resources from various clearinghouse websites. parents focusing on trust, possibly through home visits by practitioners, is effective202. Finally, removing barriers to Effectively engaging with parents not only requires participation such as inadequate timing, lack of interest, choosing among alternative programs, but also recruiting or stigmatization is as important as providing catch- parents and keeping them engaged. The most challenging up sessions to prevent children from dropping out. part of engaging with parents is probably not to choose 195 Durlak et al. (2010); Hirsch et al. (2010); Yohalem and Wilson-Ahlstrom (2010). 196 WHO (2019). 197 UNESCO and UN Women (2016). 198 WHO et al. (2016, 2018). 199 Axford et al. (2012). 200 McDonald et al. (2012). 201 Baker et al. (2011). 202 Caspe and Lopez (2006); Davidson and Campbell (2007). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 64 WORKING WITH COMMUNITIES ON SOCIAL behavior. By the end of the three-year campaign, through NORMS AND SAFE PASSAGE TO SCHOOLS television, radio and print, online multimedia campaign, educational materials and travelling video vans, over As schools do not operate in isolation, it is important to 130 million people had been reached. In addition, more engage with communities to shift norms conducive of than 75,000 rights advocates were trained to become violence, be they the endorsement of corporal punishment agents of change. The efforts resulted in a 49 percent as a way to discipline a child, gender inequality, or the use increase in the number of people aware of the Protection of violence to solve conflicts or exert power over others. of Women from Domestic Violence Act in India, and a Apart from the Good School Toolkit mentioned earlier, 15 percent increase in access to services for survivors. SASA! is a good example that shows how norms can be challenged even in a relatively short period of time. Engaging with community is also important as violence SASA! means “Now!” in Kiswahili. The program was commonly happens on the way to or from school. Ensuring developed by Raising Voices to prevent violence against a safe passage to schools usually implies identifying the women and has been implemented in Uganda by the main routes to schools, hot spots where children (both Center for Domestic Violence Prevention. It employs girls and boys of different ages) feel most vulnerable, and multiple strategies to build a critical mass of engaged placing adult monitors on those strategic routes during community members, leaders, and institutions, including commute times. Safe passages can also be enhanced, local activism, media and advocacy, communication for example by cleaning up routes, improving the materials, and training. The Activist Kit that is central to physical environment in and outside the school, making SASA! community engagement and mobilization involves sure schools can close, working with street vendors or four phases: Start, Awareness, Support, and Action. The other shop keepers in the surrounding of the schools to content evolves with each phase, with power as a central enhance positive natural surveillance, ensuring public theme. Results from a randomized controlled trial suggest lighting works, and improving road safety205 . Evaluation positive effects after three years of programming. In of the Chicago Safe Passages suggests positive results comparison to control communities, SASA! communities in terms of crime reduction on the main routes to reported a reduction in levels of violence against women school as compared to neighboring roads where the of 52 percent; an increase in the share of women and program was not implemented206 . Other examples of men who believe it is acceptable for women to refuse safe passage mechanisms include Walking Buses in Iraq sex of 28 percent; and an increase of 50 percent in with two trained adults escorting girls to school207 and the share of men and women who believe that physical the Jamaican Integrated Community Development violence against a partner is unacceptable. Essentially, Project that combined road wardens, community SASA! works with key stakeholders at the community mobilization, training in violence prevention and road level to deconstruct power in intimate partnerships203 . safety and proper signals on the main roads to school208 .   Another interesting program is the Bell Bajao! (Ring STRATEGIC APPROACHES FOR the Bell) Campaign204 . The campaign was launched in VIOLENCE PREVENTION 2008 to call on men and boys across India to take a stand against domestic violence by performing a simple Efforts to prevent VIAS can take place in a single school, bystander intervention – ringing the doorbell when or a group of schools, but ideally they should be broader they witnessed domestic violence taking place. The and led by Ministries of Education at the national level, campaign’s integrated cultural, organizational and media often in collaboration with other Ministries or agencies. strategy sought to make the issue part of mainstream To that end, systemic approaches are needed since of conversations, increase knowledge about and change the aims is to transform societies so that VIAS is no community attitudes towards domestic violence and longer accepted, justified and practiced. To sustainably towards HIV-positive women; and alter individual shift norms, parent associations and teacher unions, as 203 Michau et al. (2008); Abramsky et al. (2014). 204 Breakthrough (2013); Michau et al. (2016). 205 This is often referred to Crime Prevention Through Environmental Design. 206 Curran (2019). 207 UNICEF (2010). 208 World Bank (2019). 65 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 well as religious groups and even political parties, need groups. All stakeholders, including teacher unions, to participate and be heard. Several useful guides and political parties, religious groups, parent associations, handbooks exist in that respect, for example on engaging and children need to better understand children’s rights with religious leaders to end VIAS209. More generally, and the detrimental effects of VIAS as well as the link four steps in the strategic process can be suggested, between violence against women and violence against whether this is done nationally or at a sub-national level. children. Educational campaigns also need to focus on the benefits of non-violent alternatives as children who A first step is to set clear standards for all, which can be are not afraid in schools have a higher connectedness, done by passing laws that prohibit all forms of violence, mental health and educational achievements. and in particular violence against children in school. As mentioned earlier, on corporal punishment, according Some countries have adopted codes of conduct focusing to the Global Initiative to End All Corporal Punishment on school communities. Like legislation, codes of conduct of Children, 132 countries have prohibited the practice are useful tools to set standards and signal priorities and in schools and 56 have done so in all settings including values. They set guidelines and ethical standards for school at home. However, thirty years after the adoption of staff (teachers and principals) as well as students and the UN Convention on the Rights of the Child, corporal parents and clearly outline unacceptable behaviors such punishment in schools is still lawful in 67 countries. as the use of any type of violence, sexual harassment, Legislation is necessary but not sufficient however, and abuse of power in school. To be effective though, especially in contexts where corporal punishment is they need to be widely known and enforceable. seen as a necessity to properly discipline a child and enforcement systems are unlikely to exist or work even A second step is the development of solid diagnostics when corporal punishment is outlawed. Legislative efforts to guide the elaboration of action plans at different must be accompanied by continued and multi-pronged levels. To prevent violence and be able to target efforts awareness raising efforts. Such efforts need to happen and monitor progress, it is crucial to understand the at all levels of society and be relayed by influential prevalence of different types of violence in a given 209 Dodd (2011). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 66 context. It is also important to analyze risk and protective ensure a positive learning environment. As mentioned factors associated with different forms of violence. earlier, a whole school approach seeks to change norms Through the Safe to Learn campaign, a diagnostic tool at the school level by engaging all stakeholders, from has been developed that covers key elements that need students to teachers and principals, and parents as well as to be in place to ensure a safe learning environment community members and leaders. In implementing such for children. The tool helps identify assets and gaps in approaches, due attention should be paid to what the terms of legislation, policy, budgeting, capacity, from empirical evidence suggests in terms of what works and the national level all the way down to the school level. what may not, taking into account the form of violence Each layer of influence is looked at, with suggestions being considered, age group affected, gender dynamics, based on the evidence and best practices available. and context. This is the case for the prevention of violence as well as for steps that are taken when violence occurs. A third step is to develop a common vision and action plan not only to clearly set priorities and responsibilities, but also Finally, it must be recognized that investment projects define standards and accountability mechanisms across in the social sectors including education can exacerbate agencies and sectors. Action plans can be set at all levels of the risk of Sexual Exploitation and Abuse and Sexual influence to be closer to the context. These plans ideally are Harassment (SEA/SH). Donors and government agencies co-created, that is, they adopt a participatory process and must therefore be careful go assess such risks in their involve all relevant stakeholders, including teachers, unions, projects. At the World Bank, within the context of the parents, and community leaders. Action plans need to have Environmental and Social Framework, principles and relevant indicators on behavioral change (not just outputs) approaches to identify and mitigate the risk of SEA/ to assess progress and improve programs as needed. SH in projects are outlined in a good practice note210 currently under development, the principles of which A fourth step is to promote a whole school approach may also be of interest to government Ministries. to enhance students’ connectedness with schools, and 210 World Bank (2021). 67 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 CHAPTER 8 as illustrated in this chapter with a case study for Uganda, implementation costs may be very different for an ILLUSTRATIVE COST- initial pilot intervention in comparison to the costs when interventions are scaled up. Accounting for benefits is BENEFIT ANALYSES also fraught with potential issues. One issue relates to the economic valuation of benefits. But another issue relates The analysis so far has provided evidence of the large to the type of benefits included. Some interventions negative potential impacts of violence in school, some may have been documented to have impacts in a wide of the costs generated by violence in schools especially range of areas, while for other interventions, only a few through lower educational attainment and its implication impacts may be available. This may distort comparisons for human capital wealth, and some of the interventions of benefits due to differences in the comprehensiveness that have been proven to reduce violence in schools. of the benefits being included in the analysis. The question considered in this chapter is whether these interventions generate more economic benefits than When comparing the streams of costs (during the cost of implementing them. To show that this is implementation) with benefits (during and after often the case, this chapter provides illustrative results implementation, including in adulthood for children who from cost-benefit analyses building on the existing benefited from interventions), the issue of which discount literature. The chapter thereby helps to demonstrate rate (given the need to compute the present value of that implementing interventions to reduce violence benefits that may arise many years in the future) to use in and around schools is not only the right thing to may be debated. A higher discount rate, which may be do, but also a smart economic investment. It should warranted in developing country contexts, may reverse however be noted that many of the analyses have been the conclusion that would result from a lower discount conducted in developed countries and especially in the rate, in that the sign of the net present value estimated United States – so they may not be representative may change. This is also an additional reason why it is of the costs/benefits in developing countries. often difficult – and not advisable, to narrowly compare cost-benefit analyses conducted separately for different PRINCIPLES AND LIMITATIONS OF interventions since these analyses may have used different COST-BENEFIT ANALYSIS hypotheses, including for discount rates. Readers should not infer that one intervention is necessarily better than Cost-benefit analysis aims to compare the benefits of another simply because the benefit to cost ratio of the first interventions to their cost. When the ratio of benefits intervention appears to be higher than that of the second. to cost is high, which typically also implies high internal rates of return, an intervention may be especially Still another issue is that of the external validity of an attractive for policy makers in contexts where budget impact evaluation, and therefore of any cost-benefit resources are limited. Extensive data and assumptions analysis based on the results of the evaluation. An are typically needed to conduct a cost-benefit analysis. intervention may have been successful in a particular This implies that results can be sensitive to assumptions context of setting. This does not imply that it will as well as other parameters related to data quality. Cost- remain successful when scaled up in a country, or when benefit analyses must rely on the results of impacts implemented in a different context, or even a different evaluations since the benefits relate to the impacts of the country. It is often the case that intervention that were interventions. Such impact evaluations may be limited successful when piloted turned out not to have the in number and may themselves be more or less robust. same impacts when scaled up, or when implemented in a different context. Many of the available cost-benefit Detailed estimates of costs and benefits are also required. analyses for programs aiming to reduce VIAS have While some interventions may provide a comprehensive been conducted in Western countries, and especially accounting of all costs involved, this is often not the in the United States. It could be that the ratios of case. In the literature, information on implementation benefits to costs that could be observed for similar costs is often lacking even when information on the interventions in developing countries could be different. impacts of an intervention may be available. In addition, JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 68 For all of those reasons, the analysis provided in this interventions. In a framework on priorities for ECD, chapter is tentative and illustrative, as opposed to 25 interventions were considered as essential214 . They definitive. This being acknowledged, the chapter provides were organized according to the time period in the life illustrative results from cost-benefit analyses following of children to which they apply and grouped into five the life cycle approach used in Chapter 6. Interventions packages: pregnancy, birth, child health, preschool, and related to early childhood development (ECD) are first family support packages. Most of these interventions discussed, before considering programs implemented at have high rates of returns, with economic benefits often the primary and secondary levels specifically to reduce estimated at multiples of the cost of the interventions. the risk of violence. Thereafter, two additional sections Because of these high benefit-cost ratios, investments consider the potential for reducing costs when scaling in ECD are often considered as having higher economic up a program (using cost data from Uganda’s Good rates of returns than investments made later in life, School Toolkit as an example), and the specific case of for example, in school or in an adult’s working life215 . interventions to reducing gender-based violence. The evidence on the economic benefits from ECD EARLY CHILDHOOD INTERVENTIONS interventions is broad. Multiple interventions in a variety of contexts and countries have been shown to generate Risk factors for future violence start during pregnancy positive impacts. When the cost of implementing and the early years of a child’s life. Experiencing violence interventions is very low, as is the case for some health at home is associated with a higher risk of perpetration and nutrition interventions, the benefit-cost ratios can or victimization later in life. In addition, socio-economic be especially high. But even for types of interventions disadvantage in the early years, including poor nutrition, that are more costly to implement, such as ensuring that may affect a child’s brain, leading to negative effects in young children have the opportunity to benefit from adulthood. The first 1,000 days are especially crucial. preschools in order to be ready to start primary school, Hundreds of new neural connections are formed in much of the literature suggests high benefit to cost the brain every second in a child’s early years. Sensory ratios. At the same time, there is a danger in assuming pathways for basic vision and hearing develop first, followed that ECD interventions are always going to generate by early language skills and higher cognitive functions211 . important benefits. This is not necessarily the case. In particular, toxic stress must be avoided. As mentioned earlier, it refers to a child’s response when experiencing Consider the case of preschools and center-based ECD repeated adversity, including physical and emotional abuse programs, for which many of the more in-depth cost- as well as economic hardship and exposure to violence benefit analyses have been conducted in the United States in the household (such as intimate partner violence). and other Western countries. A summary of those analyses As is the case for stunting, toxic stress may affect brain is provided in a study216 that considers six programs: (1) development with implications in multiple areas throughout High-Scope Perry Preschool; (2) Carolina Abecedarian; adult life212. These effects are in turn associated with higher (3) Chicago Child-Parent Center Program; (4) Even Start risks of violence, whether as a victim or perpetrator. Family Literacy Program; (5) Early Head Start; and (6) Sure Start. Many of the programs also included home Ensuring that conditions for ECD are optimal is important visiting, parenting advice, health and nutrition services, to reduce the risk of violence later in life. The good news and referrals for social services. The available cost-benefit is that interventions at an early age can make a major analyses considered various types of benefits, including difference. Research suggests the need for holistic those related to education, health, employment and approaches to ECD since a child’s physical and intellectual earnings, and crime and social welfare. For five of the six well-being as well as their socio-emotional and cognitive programs, education outcomes are reported, while for development are interrelated213 . This implies that multiple three of them, impacts on crime have been estimated. conditions must be in place, which may require multiple 211 Nelson (2000). 212 See for example Aizer (2011). 213 Shonkoff, et al. (2012). 214 Denboba et al. (2014). 215 Carneiro and Heckman (2003); Heckman and Masterov (2007). 216 Dalziel et al. (2015). 69 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Table 8.1 provides a summary of the results. The first measured. By contrast, some of the more recent programs three programs were implemented early on. They had simply cannot include such benefits in the analysis. For high positive returns on investment and benefit to cost example, the Sure Start cost-benefit analysis is based ratios ranging from 2.04 to 16.14, often with multiple on outcomes observed at five years of age, thus not evaluations for the same program available which suggests including some of the large benefits from a reduction robustness in the assessment of their value. The three in crime reduction, better educational attainment, and programs implemented later with wider rollouts had higher earnings that occur later than five years after negative net present values – their benefits were smaller the program. Across the cost-benefit analyses, the than the cost of implementation, although in some cases highest net present values were observed for programs this is based on a single study. Why such large differences for which outcomes could be measured into adulthood are observed between the earlier and the more recent such as the Perry Preschool. By that reasoning, it could interventions is not fully clear as multiple factors may be still be argued that the programs for which outcomes at play in leading to diverging results for the cost-benefit could be measured show larger benefits than costs. analyses. It could be that scaling up the programs led to issues such as weaker participation from communities The key message from Table 8.1 is not that center- that are often key for program success. It could also be based and preschool interventions are not appropriate that as access to preschools became more widespread, interventions. It is rather that (1) quality implementation is at least some children in control groups benefited from essential when scaling up programs (external validity issue); similar investments, thus leading to smaller differences and (2) results from a cost-benefit analysis depend on the in outcomes between treatment and control groups. types of benefits included, which also depends on how long after an intervention an evaluation is implemented. one But perhaps the most important factor leading to cannot simply assume that programs will generate benefits potential differences in net present values relates to the that exceed costs. More generally, it is always prudent number of benefits included in the cost-benefit analyses. to acknowledge that Interventions that were successful For earlier programs, participants are now older, so in particular contexts may not work in other contexts. that a larger number of benefits into adulthood can be JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 70 Table 8.1: Cost-Benefit Analysis for Selected Preschool and ECD Programs Program Description Cost in Benefit in Net Present Ratio (2)/(1) 2011 USD (1) 2011 USD (2) Value (2)-(1) Early and small scale programs with relatively comprehensive accounting of long-term benefits Program of 21.6 months High-Scope for children 2-4 year old $19,867 to $45,789 to 2.04 to $23,468 to Perry Preschool with emphasis on cognitive $29,836 $320,702 16.14 $300,835 (United States) development Program of up to 60 Carolina months for children 6-12 Abecedarian weeks old with emphasis $45,189 $170,788 3.78 $125,599 (United States) on language and social development Program of 19.2 months for Chicago Child- children 3-4 year old with Parent Center $42,684 to 4.82 to $33,829 emphasis on cognitive and $8,855 to $9,304 Program $100,520 10.83 to $91,242 social development as well (United States) as family support More recent and larger programs typically without full accounting of long-term benefits Program of 10 months for Even Start Family children 0-8 year old with Literacy Program emphasis on improving $5,981 $0 NA −$5,981 (United States) academic achievement, especially reading Program of 22 months for children below one year old Early Head Start with emphasis on enhancing $25,796 $5,865 0.23 −$19,930 (United States) child development and strengthening families Program of variable length for children up to 4 year Sure Start $452 0.057 to −$7,421 to old with emphasis on 3arly $7,873 (England) to $902 0.115 −$6,971 education, child care, and health services Source: Adapted from Dalziel et al. (2015). SOCIO-EMOTIONAL SKILLS AND in the labor market220 . In secondary schools, for at risk RELATED PROGRAMS adolescents, approaches such as cognitive behavioral training have proven effective (Life Skills Training was In primary schools, one of the main recommendations mentioned as an example). Specific interventions can to reduce the prevalence of violence in Chapter 6 was also be implemented to prevent dating violence, as is the to implement programs helping children improve their case with the Fourth R program that aims to empower social and emotional skills. This includes competencies adolescents to build and maintain healthy relationships221 . such as self-awareness, self-management, social Finally, after-school programs that combine recreational awareness, relationship skills, and responsible decision- activities and academic support may also reduce violence. making217. Acquisition of socio-emotional skills often leads to gains in well-being, positive attitudes, and pro- How do the gains suggested by impact evaluations social behavior, reductions in risky behaviors such as translate in terms of the comparison of the resulting violence and substance abuse218 , and improvements in economic benefits with the cost of implementing the academic performance219, all of which can lead to success programs? Analysis is not available for most programs, 217 Weissberg et al. (2015). 218 Durlak et al. (2010); Collaborative for Academic, Social, and Emotional Learning (2013). 219 Durlak et al. (2011); Sklad et al. (2012). 220 Heckman and Kautz (2012). 221 Wolfe et al. (2009). 71 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 but a recent synthesis of cost-benefit analyses of SEL aggression (physical, verbal or psychological) over a interventions covers six different programs222. The synthesis prolonged period of time among peers who have an acknowledges previous cost-benefit analysis for several of imbalance of power. Reviews also suggest that intensive these programs223 , but notes that because previous analyses and long-lasting program are needed to change behaviors, often relied on different methodologies, conducting a with parental sessions contributing to the success of new cost-benefit analysis with the same methodology the programs226 . A meta-analysis of multiple evaluations across programs could provide additional insights. of anti-bullying programs suggests that the programs reduced the prevalence of bullying by about a fifth The first program is the 4Rs Program (Reading, Writing, on average227, with comprehensive programs inspired Respect, and Resolution) implemented in grades K-5 to by the Olweus model working particularly well. promote pro-social behavior and help students develop cooperative problem-solving skills. An evaluation in New As for SEL-related programs, cost-benefit analysis York City suggested a reduction a range of aggressive has been conducted for several programs aiming to behaviors as well as gains in achievement for mathematics reduce bullying. Two frequently cited program are the and reading224 . The second program, Positive Action, aims Olweus Bullying Prevention Program (OBPP), which to promote positive thinking, actions, and self-concept was mentioned in Chapter 6, as well as the KiVa anti- for all students in grades 3 to 8. The third program, Life bullying program, which includes actions targeting all Skills Training, is a classroom intervention for grades students (curriculum including student lessons and 6 to 12 to reduce substance abuse and violence225 . The online games to prevent bullying), and indicated actions fourth program, Second Step, is based on a social skills to be used when a bullying case has emerged – those curriculum to improve problem-solving and emotional actions specifically target children and adolescents management for children from pre-kindergarten to who have been involved in bullying as perpetrators or grade 10. The next program, Responsive Classroom, victims. A cost-benefit analysis of the potential benefits targets students in grades 3 to five to improve socio- from OBPP in Pennsylvania suggests that the cost of emotional skills by improving teacher efficacy. The last implementing the program would be at US$25.8 million program, Social and Emotional Training, was implemented or an average of US$7.70 per student per year. However, in Sweden) for grades 1 to 9. While some programs when start-up costs are not considered, the on-going target students at risk, many cover all children. cost of implementation falls to US$2.07 per student per year. The analysis suggests that net savings from the Table 8.2 provides results for baseline estimates. Each program through lower healthcare costs thanks to the of the six interventions generates higher benefits than expected reduction in bullying are valued at $12.30 per costs according to the review, with benefit-cost ratio student year228 . This in turn suggests a benefit-cost ratio ranging from 3.46 to 13.91 across interventions in the of 6.94 not including start-up costs (these costs would baseline scenarios. For each intervention, the authors be spread over time over multiple years, so the program also consider alternative scenarios that affect the benefits appears cost effective; in addition, other potential benefits to costs ratios, but in no case are the ratios negative. not included in the analysis could raise the benefit to This suggests that investments in SEL and related cost ratio). Results are summarized in Table 8.3. programs could generate substantial economic returns. Table 8.3 also provides data on three cost-benefit ANTI-BULLYING PROGRAMS analyses for the KiVa anti-bullying program for the Netherlands229, Wales230 , and Sweden. The analyses for In secondary schools, another recommendation in the Netherlands and Wales suggest positive net present Chapter 6 was to focus on programs that reduce the values, and as a result benefit-cost ratios well above 1. prevalence of bullying. Bullying is defined as repeated The lower ratio for the program in Wales is likely due 222 Belfield et al. (2015). 223 See especially Lee et al. (2012); Jones et al. (2008); and Miller and Hendrie (2008). 224 Jones et al. (2011). 225 Hawkins et al. (1998). 226 Farrington and Ttofi (2009).. 227 Farrington and Ttofi (2009). 228 Highmark Foundation (2018). 229 Huitsing et al. (2019). 230 McDaid (2017). JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 72 in part to a smaller set of benefits taken into account. One last but important consideration when implementing In Sweden, the analysis is presented in a different way, programs and assessing costs is whether scaling up helps showing the estimated cost per QUALY (Quality-adjusted in reducing unit costs per students. To illustrate how this life year) at €13,823. This is well below the accepted can be the case, consider the Good School Toolkit (GST) norm for cost effective interventions in the country, at in Uganda mentioned in Chapter 7. A cost-effectiveness about €50,000, suggesting that in comparison to the analysis of the program231 suggests that the cost of the norm, the intervention is cost effective (the value of program when implemented at small scale was up to the benefit to cost ratio here is computed differently, US$15 per child. Scaling up nationally could dramatically and defined simply as the norm divided by the cost, or reduce this unit cost to just a few dollars per student €50,000/€13,823, so this estimate is not comparable thanks to economies of scale in design and training costs. to the other benefit-cost ratios provided, but again, as mentioned in the introduction, one should be careful not to compare benefit-cost ratios across programs unless exactly the same methodology has been used for the various cost-benefit analyses, which is often not the case). REDUCING UNIT COSTS BY SCALING UP These illustrative cost-benefit analyses suggest that interventions in early childhood, primary schools, and secondary schools that have the potential to reduce violence in schools are often economically beneficial, in that they generate relatively high benefit to cost ratios. While these ratios are sensitive to the assumptions used in the analyses, the results suggest that reducing VIAS is a smart economic investment. Unfortunately, most of the available data are from developed countries, so that how this information translates to developing countries, and especially low income countries, is not fully clear, even if one can presume that similar results would apply. Table 8.3: Cost-Benefit Analysis for the Olweus and KiVa Anti-bullying Programs Program Description Benefit in Net Present Cost in 2015-16 (1) Ratio (2)/(1) 2015-16 (2) Value (2)-(1) Olweus Olweus curriculum $2.07 6.94 $14.37 (*) $12.30 (United States) in K-12 schools (on-going only) (on-going only) KiVa KiVa curriculum in €203 €819 to €1,363 4.04 to 6.72 €203 to €818 (Netherlands) primary schools (4 years) KiVa KiVa curriculum in £656 £1,037 1.58 £381 (Wales) primary schools (4 years) €13,823 per KiVa KiVa curriculum in €829 QUALY vs. 3.62 (**) NA (Sweden) primary schools (4 years) €50,000 norm Source: Compiled by the authors. Note: (*) The value of $14.37 is not provided by the authors but reconstructed by adding the on-going cost of the program and the net benefit. (**) The ratio of benefits to costs is computed differently, as the accepted norm in the country for acceptable cost per QUALY divided by the estimated cost per QUALY for the program. 231 Greco (2018). 73 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 REFERENCES Abramovay, M., and Das Graças Rua, M. (2002). Violence in Schools. Brasilia: UNESCO. Abramsky, T., Devries, K., Kiss, L., Nakuti, J., Kyegombe, N., Starmann, E., Cundill, B., Francisco, L., Kaye, D., Musuya, T., Michau, L., & Watts, C. (2014). Findings from the SASA! Study: A cluster randomized controlled trial to assess the impact of a community mobilization intervention to prevent violence against women and reduce HIV risk in Kampala, Uganda. BMC Medicine. 12. 10.1186/s12916-014-0122-5. Alcaraz, R. (2002). School Violence. Fact Sheet. Southern California Center of Excellence on Youth Violence Prevention, Riverside, CA: University of California Riverside. Alda, E. (2005). School-Based Violence Prevention. Technical Brief 11. Washington DC: Inter-American Development Bank. Allen, M., Witt, P. L., Wheeless, L. R., Allen, M., Witt, P. L., & Wheeless, L. R. (2007). The role of teacher immediacy as a motivational factor in student learning: Using meta-analysis to test a causal model. Communication Education, 55(1), 21–31. Arcia, E. (2006). Achievement and enrollment status of suspended students: outcomes in a large, multicultural school district. Education and Urban Society, 38(3), 359-369. Asscher, J. J., Dekovic, M., Manders, W., van der Laan, P. H., Prins, P. J. M., & van Arum, S. (2014). Sustainability of the effects of multisystemic therapy for juvenile delinquents in the Netherlands: effects on delinquency and recidivism. Journal of Experimental Criminology, 10, 227–243. Axford, N., Lehtonen, M., Kaoukji, D., Tobin, K. & Berry, V. (2012). Engaging parents in parenting programs: Lessons from research and practice. Children and Youth Services Review, 34, 2061–2071. Baafi, A., & R. Kwabena (2020). Teacher-Student Relationship And Student Learning Outcomes In Senior Public Secondary Schools In Ghana. European Journal of Education Studies, 6(12), 147-161. Baker, C.N., Arnold, D.H., & Meagher, S. (2011). Enrollment and attendance in a parent training prevention program for conduct problems. Prevention Science, 12(2), 126-138. Baker-Henningham, H., Bowers, M., Francis, T. et al. (2017). The Irie Classroom Toolbox: a cluster randomised trial of a universal violence prevention programme in Jamaican preschools. Abstract submitted to the Society for Prevention Research Annual Meeting. Bair-Merritt, M. H., Blackstone, M. & Feudtner, C. (2006). Physical health outcomes of childhood exposure to intimate partner violence: a systematic review. Review Pediatrics, 117(2), e278-90. Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. New York: Prentice-Hall. Belfield, C., Klapp, A., & Lebvin, H. M. (2015). The Economic Value of Social and Emotional Learning, Journal of Benefit-Cost Analysis, 6(03). Black, M. M., S. P. Walker, L. C. H. Fernald, C. T. Andersen, et al. (2017). Early Child Development Coming of Age: Science through the Life-Course, The Lancet, 389 (10064): 77-90. Blum, R. W., & Rinehart, P. (1997). Reducing the Risk: Connections That Make a Difference in the Lives of Youth. Minneapolis: University of Minnesota, Division of General Pediatrics, Adolescent Health. Blum, R. W., Beuhring, T., & Rinehart, P. M. (2000). Protecting teens: Beyond race, income and family structure. Minneapolis: University of Minnesota, Center for Adolescent Health. Botea, I., S. Chakravarty, and S. Haddock, and Q. Wodon. 2017. Interventions Improving Sexual and Reproductive Health Outcomes and Delaying Child Marriage and Childbearing for Adolescent Girls. Ending Child Marriage Notes Series. Washington, DC: The World Bank. Bradshaw, C. P., Waasdorp, T. E., & Leaf, P. J. (2012). Effects of school-wide positive behavioral interventions and supports on child behavior problems. Pediatrics, 130(5), e1136–e1145. Breakthrough Insights. 2013. Breakthrough’s Bell Bajao! A Campaign to Bring Domestic Violence to a Halt. Available at: http://www.breakthrough. tv/o/wp-content/files_mf/1330816837BellBajao_Insight.pdf. Brotman, L.M., Dawson-McClure, S., Calzada, E.J., Huang, K.Y., Kamboukos, D., Palamar, J.J., & Petkova, E. (2013) Cluster (school) RCT of ParentCorps: impact on kindergarten academic achievement. Pediatrics, 131(5):e1521-9. Brotman, L. M., Dawson-McClure, S., Kamboukos, D., Huang, K. Y., Calzada, E. J., Goldfeld, K., & Petkova, E. (2016). Effects of ParentCorps in Prekindergarten on Child Mental Health and Academic Performance: Follow-up of a Randomized Clinical Trial Through 8 Years of Age. JAMA pediatrics, 170(12), 1149–1155. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 74 Botvin, G. J., K. W. Griffin, & T. R. Nichols. (2006). Preventing Youth Violence and Delinquency through a Universal School-Based Prevention Approach. Prevention Science, 7: 403–8. Carneiro, P., & D. Evans. (2013). The Impact of Formal Child Care Attendance in Rio de Janeiro on Child Development and Maternal Outcomes. Working paper. Caspe, M. & Lopez, E. (2006). Lessons from Family-Strengthening Interventions: Learning from Evidence-Based Practice. Harvard Family Research Project. Cahu, P. & Quota, M. (2019). Does School Safety and Classroom Disciplinary Climate Hinder Learning? Evidence from the MENA Region. Policy Research Working Paper No. 8822. Washington, DC: The World Bank. Catalano, R.F., Oxford, M.L., Harachi, T.W., Abbott, R.D., & Haggerty, K.P. (1999). A test of the social development model to predict problem behaviour during the elementary school period. Criminal Behaviour and Mental Health, 9:39–56. Chioda, L. (2017). Stop the Violence in Latin America: A Look at Prevention from Cradle to Adulthood. Latin American Development Forum. Washington, DC: The World Bank. Cornell, D. G., Allen, K., & Fan, X. (2012). A randomized controlled study of the Virginia student threat assessment guidelines in kindergarten through grade 12. School Psychology Review, 41(1), 100–115. Cornelius-white, J. (2007). Learner-centered teacher-student relationships are effective: a meta-analysis. Review of Educational Research, 77(1), 113–143. Curran, F. C. (2019). Does the Chicago Safe Passage Program Reduce Reported Crime Around Elementary Schools? Evidence From Longitudinal, Geocoded Crime Data. Criminal Justice Policy Review, 30(9), 1385–1407. Currie, J., & Tekin, E. (2012). Understanding the cycle: childhood maltreatment and future crime. The Journal of human resources, 47(2), 509–549. Dalziel, K. M., Halliday, D. & Segal, L. (2015). Assessment of the Cost–Benefit Literature on Early Childhood Education for Vulnerable Children: What the Findings Mean for Policy. SAGE Open, January-March, 1–14 Davidson, G. & Campbell, J. (2007). An Examination of the Use of Coercion by Assertive Outreach and Community Mental Health Teams in Northern Ireland, British Journal of Social Work, 37(3) 537-555. Dawson-McClure, S., Calzada, E., Huang, K.Y., Kamboukos, D., Rhule, D., Kolawole, B., Petkova, E., & Brotman, L.M. (2015). A population- level approach to promoting healthy child development and school success in low-income, urban neighborhoods: impact on parenting and child conduct problems. Prev Sci. 16(2):279-90. Deković, M., Asscher, J. J., Manders, W. A., Prins, P. J., & van der Laan, P. (2012). Within-intervention change: mediators of intervention effects during multisystemic therapy. Journal of consulting and clinical psychology, 80(4), 574–587. Deloitte Access Economics. (2019). The economic cost of violence against children and young people. Deloitte Access Economics. Denboba, A. D.; Sayre, R.; Wodon, Q. et al. (2014). Stepping Up Early Childhood Development : Investing in Young Children for High Returns. Washington, DC: the World Bank. Devries, K. et al. (2015). The Good School Toolkit for reducing physical violence from school staff to primary school students: a cluster- randomised controlled trial in Uganda. Lancet Global Health, 385, e378–86. Devries, K., Fabbri, C., Allen, E., Barongo, V. et al. (2019). Preventing violence against children in schools (PVACS): Protocol for a cluster randomised controlled trial of the EmpaTeach behavioural intervention in Nyarugusu refugee camp. BMC Public Health. 19. 10.1186/s12889- 019-7627-y. Durlak, J. A., Weissberg, R. P. & Pachan, M. (2010). A Meta-Analysis of After-School Programs That Seek to Promote Personal and Social Skills in Children and Adolescents. American Journal of Community Psychology, 45(3-4):294-309. Durrant, J. E. (1999). Evaluating the success of Sweden’s corporal punishment ban. Child Abuse & Neglect, 23(5), 435–448. Fang, X. (2015). The Economic Burden of the Health Consequences of Violence Against Children in Cambodia. Cambodia, Phnom Penh: Ministry of Women’s Affairs and UNICEF. Fang, X., Brown, D. S., Florence, C. S., & Mercy, J. A. (2012). The economic burden of child maltreatment in the United States and implications for prevention. Child Abuse & Neglect, 36(2), 156-165. Fang, X., Fry, D. H., Brown, D. S. et al. (2015). The burden of child maltreatment in the East Asia and Pacific region. Review Child Abuse Negl., 42,146-62. 75 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Fang, X., Fry, D. A., Ji. K. et al. (2015). The burden of child maltreatment in China: a systematic review. Bulletin of the World Health Organisation 93(3), 176-185C. Fang, X., Zheng, X. Fry, D. A. et al. (2017). The Economic Burden of Violence against Children in South Africa. Int J Environ Res Public Health, 14(11): 1431. Fargo, J. D. (2009). Pathways to Adult Sexual Revictimization: Direct and Indirect Behavioral Risk Factors Across the Lifespan. Journal of Interpersonal Violence, 11(24). Farrington, D. P., & Welsh, B. C. (2007). Saving Children from a Life of Crime: Early Risk Factors and Effective Interventions. New York: Oxford University Press. Farrington, D. & Ttofi, M. (2009). School-Based Programs to Reduce Bullying and Victimization. Campbell Systematic Reviews 2009:6. FAWE. 2018. Gender Responsive Pedagogy: A Toolkit for Teachers and Schools. Second updated edition. Nairobi: Forum for African Women Educationalists. Fearon, J. & Hoeffler, A. (2014). Benefits and Costs of the Conflict and Violence Targets for the Post-2015 Development Agenda: Post-2015 Consensus. Conflict and Violence Assessment Paper. Copenhagen: Copenhagen Consensus Center. Field, E. & Ambrus, A. (2008). Early Marriage, Age of Menarche, and Female Schooling Attainment in Bangladesh. Journal of Political Economy, 116(5), 881-930. Finkelhor, D., Shattuck, A., Turner, H. A., Ormrod, R. & Hamby, S. L. (2011). Polyvictimization in Developmental Context. Journal of Child & Adolescent Trauma, 4, 291–300. Fisher, K. (2001). Building Better Outcomes: The Impact of School Infrastructure on Student Outcomes and Behavior. Rubida Research for the Commonwealth Department of Education, Training and Youth Affairs, Australia. Font, S. and Cage, J. (2018). Dimensions of physical punishment and their associations with children’s cognitive performance and school adjustment. Child Abuse & Neglect, 75: 29-40. Frankenberg, S. J., Holmqvist, R., & Rubenson, B. (2010). The care of corporal punishment: Conceptions of early childhood discipline strategies among parents and grandparents in a poor and urban area in Tanzania. Childhood, 17(4), 455-469. Fry, D., Fang, X., Elliott, S. et al. (2018). The relationships between violence in childhood and educational outcomes: A global systematic review and meta-analysis. Child Abuse Negl., 75, 6-28. Fulu, E., Miedema, S., Roselli, T. et al. (2017). Pathways between childhood trauma, intimate partner violence, and harsh parenting: findings from the UN Multi-country Study on Men and Violence in Asia and the Pacific. Lancet Global Health, 5(5), e512-e522. Garrard, W. M., and M. W. Lipsey. (2007). Conflict Resolution Education and Antisocial Behavior in U.S. Schools: A Meta-Analysis. Conflict Resolution Quarterly, 25(1), 9–38. Gemignani, R. & Wodon, Q. (2017). Gender Roles and Girls’ Education in Burkina Faso: A Tale of Heterogeneity between Rural Communities, American Review of Political Economy, 11(2), 163-75. Gershoff, E. T. (2002). Corporal punishment by parents and associated child behaviors and experiences: A meta-analytic and theoretical review. Psychological Bulletin, 128(4), 539–579. Gershoff, E., & Grogan-Kaylor, A. (2016). Spanking and child outcomes: Old controversies and new meta-analyses. Journal of family psychology : JFP : journal of the Division of Family Psychology of the American Psychological Association, 30(4), 453-69 . Gordon, D. A., K. Graves, and J. Arbuthnot. (1995). The Effect of Functional Family Therapy for Delinquents on Adult Criminal Behavior. Criminal Justice and Behavior, 22, 60–73. Gottfredson, D. C. (2001). Schools and Delinquency. New York: Cambridge University Press. Gottfredson, D. C., S. A. Gerstenblith, D. A. Soule, S. C. Womer, & S. Lu. (2004). Do After-School Programs Reduce Delinquency? Prevention Science, 5(4): 253–66. Greco, G., Knight, L., Ssekadde, W. et al. 2018. Economic evaluation of the Good School Toolkit: an intervention for reducing violence in primary schools in Uganda. BMJ Global Health, 3, e000526. Guedes, A., Bott, S., Garcia-Moreno, C., & Colombini, M. (2019). Bridging the gaps: a global review of intersections of violence against women and violence against children. Global Health Action, 9, 10.3402/gha.v9.31516. Guerra, N., Williams, K., Walker, I., & Meeks-Gardner, J. (2013). Building an Ecology of Peace in Jamaica: New Approaches to Understanding Youth Crime and Violence and Evaluating Prevention Strategies. Working paper. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 76 Guerrero, G. and Rojas, V. (2016). Understanding Children’s Experiences of Violence in Peru: Evidence from Young Lives, Innocenti Working Paper 2016-17. UNICEF Office of Research, Florence. Hagborg, J. M., Berglund, K., & Fahlke, C. (2018). Evidence for a relationship between child maltreatment and absenteeism among high- school students in Sweden. Child abuse & neglect, 75, 41–49. Hallinan, M T. (2008). Teacher Influences on Students’ Attachment to School. Sociology of Education, 81(3), 271-283. Hawkins, J. D., Doueck, H. J., & Lishner, D. M. (1988). Changing teaching practices in mainstream classrooms to improve bonding and behavior of low achievers. American Educational Research Journal, 25(1), 31–50. Hawkins, J. D., Von Cleve, E., Catalano, R. (1991). Reducing Early Childhood Aggression: Results of a Primary Prevention Program. Journal of the American Academy of Child & Adolescent Psychiatry, 30(2), 208-217. Hazelden Foundation. (2007). Olweus Bullying Prevention Program Scope and Sequence. Hazelden Foundation. Heckman, J., Moon, S.Y., Pinto, R., Savelyev, P., & Yavitz, A. (2010). Analyzing Social Experiments as Implemented: A Reexamination of the Evidence from the HighScope Perry Preschool Program. Quantitative Economics, 1(1), 1–46. Heise, L. 2011. What Works to Prevent Partner Violence? An Evidence Overview. Working paper. London: London School of Hygiene and Tropical Medicine. Heller, S., Pollack, H., Ander, R. & Ludwig, J. (2013). Preventing Youth Violence and Dropout: A Randomized Field Experiment. NBER Working Paper 19014, Cambridge, MA: National Bureau of Economic Research. Hemphill, S. A., & Hargreaves, J. (2010). School suspensions. A resource for teachers and school administrators. Centre for Adolescent Health, Murdoch Childrens Research Institute, Melbourne. Highmark Foundation. (2018). The Cost Benefit of Bullying Prevention: A First-time Analysis of Savings. Highmark Foundation. Hill, P. L., Roberts, B., Grogger, J., Guryan, J., & Sixkiller, K. (2011). Decreasing Delinquency, Criminal Behavior, and Recidivism by Intervening on Psychological Factors Other Than Cognitive Ability: A Review of the Intervention Literature. In Controlling Crime: Strategies and Tradeoffs, edited by P. Cook, J. Ludwig & J. McCrary. Cambridge, MA: National Bureau of Economic Research. Hillis S, Mercy J, Amobi A, et al. 2016. Global Prevalence of Past-year Violence Against Children: A Systematic Review and Minimum Estimates. Pediatrics, 137(3), e20154079 Hirsch, B. J., Mekinda, M.A. & Stawicki, J. (2010). More Than Attendance: The Importance of After-School Program Quality. Am J Community Psychol, 45, 447–452. Hoddinott, J., Behrman, J. R., Maluccio, J. A. , et al. (2013). Adult Consequences of Growth Failure in Early Childhood, American Journal of Clinical Nutrition, 98(5), 1170-1178. Hölzel, B. K., Lazar, S. W., Gard, T., Schuman-Olivier, Z., Vago, D. R., & Ott, U. (2011). How Does Mindfulness Meditation Work? Proposing Mechanisms of Action From a Conceptual and Neural Perspective. Perspectives on Psychological Science, 6(6), 537–559. Horton S., & Steckel, R. (2013). Global economic losses attributable to malnutrition 1900-2000 and projections to 2050, in B. Lomborg. Ed. The Economics of Human Challenges, Cambridge, U.K.: Cambridge University Press. Hotaling. G. T. & Sugarman, D. B. (1986). An analysis of risk markers in husband to wife violence: the current state of knowledge. Violence Vict. 1(2), 101-24. Hu, X. Z., Lipsky, R. H., Zhu, G., Akhtar, L. A., Taubman, J., Greenberg, B. D., et al. (2006). Serotonin transporter promoter gain-of-function genotypes are linked to obsessive–compulsive disorder. American Journal of Human Genetics, 78, 815−826. Huesmann, L. Rowell, Leonard D. Eron, Monroe M. Lefkowitz, & Leopold O. Walder. (1984). The Stability of Aggression over Time and Generations. Developmental Psychology, 20:1120–34. Huitsing, G., Barends, S. I., & Lokkerbol, J. (2019). Cost-benefit Analysis of the KiVa Anti-bullying Program in the Netherlands, International Journal of Bullying Prevention 2(03), 215–224. Ialongo, N., Poduska, J., Werthamer, L., & Kellam, S. (2001). The distal impact of two first-grade preventive interventions on conduct problems and disorder in early adolescence. Journal of Emotional and Behavioral Disorders, 9(3), 146–160. Institute for Economics and Peace (2014). The Economic Cost of Violence Containment. Sydney: Institute for Economics and Peace. Inter-Agency Standing Committee. (2007). IASC Guidelines on mental health and psychosocial support in emergency settings. Geneva: IASC. 77 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 International Rescue Committee. (2019). Research at the International Rescue Committee. Looking Back on 2018, Looking Forward in 2019. New York: International Rescue Committee. Jewkes, R., Levin, J., & Penn-Kekana, L. (2002). Risk factors for domestic violence: findings from a South African cross-sectional study. Soc. Sci. Med., 55(9), 1603-17. Jimenez, S., Niles, B. & Park, C. (2010). A mindfulness model of affect regulation and depressive symptoms: Positive emotions, mood regulation expectancies, and self-acceptance as regulatory mechanisms. Personality and Individual Differences, 49, 645-650. Kerr-Wilson, A., Gibbs, A., McAslan Fraser E., Ramsoomar, L., Parke, A., Khuwaja, HMA., and Jewkes, R. (2020). A rigorous global evidence review of interventions to prevent violence against women and girls. What Works to Prevent Violence Against Women and Girls Global Programme, Pretoria, South Africa. Krug, E., Dahlberg, L., Mercy, J., Zwi, A., & Lozano, R. Eds. (2002). World Report on Violence and Health. Geneva: World Health Organization. Know Violence in Childhood. 2017. Ending Violence in Childhood: Global Report 2017. New Delhi, India: Know Violence in Childhood. Krug, E. G., Dahlberg L. L., Mercy, J. A., Zwi, A. B., & Lozano, R.. Eds. 2002. World Report on Violence and Health. Geneva: World Health Organization. Lange, G. M., Wodon, Q. & Carey. K. (2018). Eds. The Changing Wealth of Nations 2018: Sustainability into the 21st Century. Washington, DC: The World Bank. Lewis, K. M., Schure, M. B., Bavarian, N., DuBois, D. L., Day, J., Ji, P., & Flay, B. R. (2013). Problem behavior and urban, low-income youth: a randomized controlled trial of positive action in Chicago. American Journal of Preventive Medicine, 44(6), 622–630. Lösel F, Farrington DP. (2012). Direct protective and buffering protective factors in the development of youth violence. Am J Prev Med., 43(2 Suppl 1), S8-S23. McCarthy, M., Taylor, P. Norman, R. E. et al. (2016). The lifetime economic and social costs of child maltreatment in Australia. Children and Youth Services Review, 71, 217-226. MacMillan, H. L., Wathen, N., Jamieson, E. et al. (2009). Screening for intimate partner violence in health care settings: a randomized trial. JAMA, 302(5):493-501. Mahoney, J., Durlak, J. & Weissberg, R. (2018). An update on social and emotional learning outcome research. Phi Delta Kappan. 100, 18-23. Masten, A. S., & Reed, M. J. (2002). Resilience in Development. In C. R. Snyder, & S. J. Lopez. Eds., Handbook of Positive Psychology. New York: Oxford University Press. Massey, A. (2011). Best behaviour: school discipline, intervention and exclusion. Policy Exchange, 75. McDaid, D., Hewlett. E., & Park, A. (2017). Understanding effective approaches to promoting mental health and preventing mental illness. OECD Health Working Paper Series No 97. Paris: OECD. McDonald, L., Coover, G., Sandler, J., Thao, T. & Shalhoub, H. (2012). Cultural adaptation of an evidence-based parenting programme with elders from South East Asia in the US: Co-producing Families and Schools Together-FAST. Journal of Children’s Services, 7, 113-127. McKeering, P. & Hwang, Y.-S. (2019). A Systematic Review of Mindfulness-Based School Interventions with Early Adolescents. Mindfulness, 10.1007/s12671-018-0998-9. Meltzer, H., Doos, L., Vostanis, P. et al. (2009). The mental health of children who witness domestic violence. Child & Family Social Work, 14(4), 491-501. Mercy, J., Rosenberg, M., Powell, K., Broome, C., & Roper W. (1993). Public health policy for preventing violence. Health Aff (Millwood), 12(4), 7-29. Michail, S. (2011). Understanding school responses to students’ challenging behaviour: a review of literature. Improving Schools, 14(2), 156–171. Michau L, Chevannes C, Hundle A, Ensor-Sekitoleko D, McMullen K, Moreaux M, & Sauvé S (2008). The SASA! Activist Kit for Preventing Violence Against Women and HIV. Raising Voices, Kampala. Michau, L., Horn, J., Bank, A., Dutt, M., & Zimmerman, C. (2015). Prevention of violence against women and girls: lessons from practice. Lancet, 385(9978), 1672–1684. Mitu, K., Ala Uddin, M. Camfield, L. & Muz, J. (2019). Adolescent Bodily Integrity and Freedom from Violence in Dhaka, Bangladesh. Policy Note, Gender and Adolescence: Global Evidence, London. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 78 Montenegro, C. E. and Q. Wodon. 2020. Measuring the Returns to Quality in Primary Education Globally Using Literacy as a Proxy, Mimeo. Washington, DC: The World Bank. Murata, K., P. Weihe, E. Budtz-Jorgensen, P. J. Jorgensen, & P. Grandjean. (2004). Delayed Brainstem Auditory Evoked Potential Latencies in 14-Year-Old Children Exposed to Methylmercury. Journal of Pediatrics, 144(2), 177–83. Napoli, M., Krech, P., & Holley, L. (2005). Mindfulness Training for Elementary School Students: The Attention Academy. Journal of Applied School Psychology, 21, 99-125. Namy, S., Carlson, C., O’Hara, K. et al. (2017). Towards a feminist understanding of intersecting violence against women and children in the family. Social Science & Medicine, 184, 40-48. Naz, A., Khan, W. Daraz, U., Hussain, M. & Khan, Q. 2011. The Impacts of Corporal Punishment on Students’ Performance/Career and Personality Development Up To Secondary Level Education in Khyber Pakhtunkhwa Pakistan. International Journal of Business and Social Science, 2(12). 130-140. Nelson, S. (2009) Care and Support Needs of Men who Survived Childhood Sexual Abuse: Report of a qualitative research project. The University of Edinburgh: Edinburgh Nguyen, M. C., & Wodon, Q. (2014). Impact of Child Marriage on Literacy and Educational Attainment in Africa, Background Paper for Fixing the Broken Promise of Education for All. Paris and New York: UNESCO Institute of Statistics and UNICEF. Noltemeyer, A. L., & Ward, R. M. (2015). Relationship between school suspension and student outcomes: a meta-analysis. School Psychology Review, 44(2), 224–240. Nove, A., Matthews, Z., Neal, S. & Camacho, A. V. (2014). Maternal mortality in adolescents compared with women of other ages: evidence from 144 countries. The Lancet Global Health, 2(3), 155-64. Oberle, E., Domitrovich, C., Meyers, D. & Weissberg, R. (2016). Establishing systemic social and emotional learning approaches in schools: a framework for schoolwide implementation. Cambridge Journal of Education, 46. 1-21. Obsuth, I., Sutherland, A., Cope, A., Pilbeam, L., Murray, A. L., & Eisner, M. (2016). London education and inclusion project (LEIP): results from a cluster-randomized controlled trial of an Intervention to reduce school exclusion and antisocial behavior. Journal of Youth & Adolescence, 46, 538–557. OECD (2015). PISA 2015 Results (Volume III): Students’ Well-Being. Paris: OECD. Office of the SRSG on Violence against Children. (2016). Tackling VIAS: A Global Perspective - Bridging the gap between standards and practice. New York: Office of the SRSG on Violence against Children. Okonofua, J. A., Pauneskua, D., & Walton, G. M. (2016). Brief intervention to encourage empathic discipline cuts suspension rates in half among adolescents. PNAS, 113(19), 5221–5226. Olds, D, L. M. Pettitt, J. Robinson, C. Henderson Jr., J. Eckenrode, H. Kitzman, R. Cole, & J. Powers. (1998). Reducing Risks for Antisocial Behavior with a Program of Pre-natal and Early Childhood Home Visitation. Journal of Community Psycholology, 26 (1): 65–83. Oliver, R. M., Wehby, J. H., & Reschly, D. J. (2011). Teacher classroom management practices: effects on disruptive or aggressive student behavior. Campbell Systematic Reviews, 4, 1–55. Onagoruwa, A. & Wodon, Q. (2018). Measuring the Impact of Child Marriage on Total Fertility: A Study for Fifteen Countries. Journal of Biosocial Science, 50(5), 626-39. Park-Higgerson, H., Perumean-Chaney, S., Bartolucci, A., Grimley, D. & Singh, K. (2008). The Evaluation of School-Based Violence Prevention Programs: A Meta-Analysis. The Journal of school health, 78, 465-79. Payet, J.-P. & Franchi, V. (2008). The Rights of the Child and The Good of the Learners: A Comparative Ethnographical Survey on the Abolition of Corporal Punishment in South African Schools. Childhood: A Global Journal of Child Research, 15. Pereznieto, P. Harper, C. Clench, B. and Coarasa, J. (2010). The Economic Impact of School Violence. London: Plan International & ODI. Pereznieto, P., Montes, A., Routier, S. & Langston, L. (2014). The costs and economic impact of violence against children. Londomn: ODI & ChildFund Alliance. Perlman, D., Adamu, F. & Wodon, Q. 2018. Why Do Adolescent Girls Drop Out of School in Niger? A Combined Quantitative and Qualitative Analysis, Marchés et Organisations, 32(2), 179-94. Promundo and Sonke Gender Justice. (2018). Breaking the Cycle of Intergenerational Violence: The Promise of Psychosocial Interventions to Address Children’s Exposure to Violence. Washington, DC: Promundo-US and Cape Town: Sonke Gender Justice. 79 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 Psacharopoulos. G., & Patrinos, H. A. (2018). Returns to Investment in Education: A Decennial Review of the Global Literature, Education Economics, 26(5), 445-458. Raising Voices and the African Women’s Development Fund. (2019). Preventing Violence against Women: A Primer for African Women’s Organisations. Kampala, Uganda. Ranieri, M. (2019). En el fuego cruzado: El impacto de la violencia de maras y pandillas en la educación en el Triángulo Norte de Centroamérica. Save the Children. Reynolds, A. J., Temple, J. A., Ou, S. R., Arteaga, I. A., & White, B. A. (2011). School-based early childhood education and age-28 well-being: effects by timing, dosage, and subgroups. Science, 333(6040), 360–364. Rijsdijk, L. E., Bos, A. E. R., Ruiter, R. A. C. et al. (2011). The World Starts with Me: a multi-level evaluation of a comprehensive sex education programme targeting adolescents in Uganda. BMC Public Health, 11: 334. Roorda, D., Koomen, H. & Oort, F. (2011). The influence of affective teacher – student relationships on students’ school engagement and achievement: a meta-analytic approach. Review of Educational Research, 81(4), 493–529. Rossman, S., and Morley, E. (1996). Introduction. Education and Urban Society, 28(4), 395–411. Saavedra, J. , McDonald, L. Quota, M. & Wodon, Q. (2021). The unacceptable and intolerable reality of corporal punishment of children and youth. Education for Global Development Blog. Washington, DC: The World Bank. Safe to Learn. (2021). Joining Forces to Embed Violence Prevention and Response in Education Systems Globally. Safe to Learn Strategy 2021-2024. Global Partnership to End Violence Against Children. Geneva, Switzerland. Schäfer, M. & Korn, S. (2002). Bullying als Gruppenphänomen. Manuskript unter Revision bei der Zeitschrift für Entwicklungspsychologie und Pädagogische Psychologie. Schonert-Reichl, K., Kitil, J., & Hanson-Peterson, J. 2017. To Reach the Students, Teach the Teachers: A National Scan of Teacher Preparation and Social & Emotional Learning. A Report Prepared for CASEL. Collaborative for Academic, Social, and Emotional Learning. Schweinhart, L. J., Barnes, H. V., & Weikart, D. P. (1993). Significant benefits: the HighScope Perry preschool study through age 27. Ypsilanti: HighScope Press. Sebastião, J. Tomas de Almeida, A., & Campos, J. (2002). A report from Portugal. In Smith, P. (org). Tackling violence in schools on a European wide basis. Sprague, J., Biglan, A., Rusby, J., Gau, J., & Vincent, C. (2017). Implementing school wide PBIS in middle schools: Results of a randomized trial. Journal of Health Science & Education, 1(2), 1-10. Timmons-Mitchell, J., Bender, M., Kishna, M. & Mitchell, C. (2006). An Independent Effectiveness Trial of Multisystemic Therapy With Juvenile Justice Youth. Journal of clinical child and adolescent psychology: the official journal for the Society of Clinical Child and Adolescent Psychology, 53(35), 227-36. Tremblay, R., & Craig, W. (1995). Developmental Crime Prevention. Crime and Justice, 19, 151-236. Tremblay, R., Daniel, S., Nagin, J., et al. (2004). Physical Aggression during Early Childhood: Trajectories and Predictors. Pediatrics 114 (1): e43–e50. UNESCO Institute of Statistics. (2017). Reducing Global Poverty through Universal Primary and Secondary Education. Policy paper 32/Fact Sheet 44. Montreal, Canada: UNESCO Institute of Statistics. UNESCO Institute of Statistics. (2019). New Methodology Shows that 258 Million Children, Adolescents and Youth Are Out of School. Fact Sheet no. 56. Montreal: UNESCO Institute of Statistics. UNESCO and UNGEI (2014). School-related gender-based violence in the Asia-pacific region. Bangkok: UNESCO. UNESCO and UN Women (2016). Global violence on addressing school-related gender-based violence. France: UNESCO. UNGEI. (2018). A Whole School Approach to Prevent School-Related Gender-Based Violence: Minimum Standards and Monitoring Framework, New York: United Nations Girls’ Education Initiative. UNICEF. (2014). Hidden in Plain Sight: A Statistical Analysis of Violence against Children. New York: UNICEF. UNICEF. (2017). A Familiar Face: Violence in the Lives of Children and Adolescents. New York: UNICEF. UNICEF. (2019). The Economic Burden of Violence against Children: Analysis of Selected Health and Education Outcomes - Nigeria case Study. New York: UNICEF. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 80 UNICEF and FIA. (2020). Guidance for Safe and Healthy Journeys to School During the Covid-19 Pandemic and Beyond. UNICEF: New Yok. USAID. (2015). What is the Cost of School-Related Gender-Based Violence? Factsheet. Washington, DC: USAID. Valdebenito, S., Eisner, M., Farrington, D.P. et al. (2019). What can we do to reduce disciplinary school exclusion? A systematic review and meta-analysis. Journal of Experimental Criminology, 15, 253–287. Violence Prevention Institute (2001). What Factors Foster Resiliency Against Violence?. School Violence prevention fact sheets No 5. Oakland, CA. Ward, B., & Gersten, R. (2013). A randomized evaluation of the safe and civil schools model for positive behavioral interventions and supports at elementary schools in a large urban school district. School Psychology Review, 42(3), 317–333. Watson, R. (1995). A guide to violence prevention. Educational Leadership, 55(2), 57–59. Weare K. (2013). Developing mindfulness with children and young people: a review of the evidence and policy context. J. Children Serv., 8 141–153. Weiss, B., Han, S., Harris, V., Catron, T., Ngo, V. K., Caron, A., Gallop, R., & Guth, C. (2013). An independent randomized clinical trial of multisystemic therapy with non-court-referred adolescents with serious conduct problems. Journal of consulting and clinical psychology, 81(6), 1027–1039. Weissberg, R. P., & Cascarino, J. (2013). Academic Learning + Social-Emotional Learning = National Priority. Phi Delta Kappan, 95(2), 8–13. Weissberg, R., Durlak, J., Domitrovich, C. & Gullotta, T.P. (2015). Social and emotional learning: Past, present, and future. Handbook for social and emotional learning: Research and practice. 3-19. Welsh, B., & Farrington, D. (2008). Effects of Improved Street Lighting on Crime. Campbell Systematic Reviews 2008: 13 (September 24). Wilkins, N. J., Zhang, X., Macka, K. A. et al. (2019). Societal determinants of violent death: The extent to which social, economic, and structural characteristics explain differences in violence across Australia, Canada, and the United States. SSM - Population Health, 8, 100431. Wodon, Q., & de la Brière, B. (2018). Unrealized Potential: The High Cost of Gender Inequality in Earnings. The Cost of Gender Inequality Notes Series. Washington, DC: The World Bank. Wodon, Q., Male, C., & Onagoruwa, A. (2020). A Simple Approach to Measuring the Share of Early Childbirths Likely Due to Child Marriage in Developing Countries. Forum for Social Economics, 49(2), 166-79. Wodon, Q., Male, C., Nayihouba, A., Onagoruwa et al. (2017). Economic Impacts of Child Marriage: Global Synthesis Report. Washington, DC: The World Bank and ICRW. Wodon, Q., Montenegro, C., Nguyen, H., and Onagoruwa, A. (2018). Missed Opportunities: The High Cost of Not Educating Girls. The Cost of Not Educating Girls Notes Series. Washington, DC: The World Bank. Wodon, Q., Nguyen, C., and Tsimpo, C. (2016). Child Marriage, Education, and Agency in Uganda, Feminist Economist, 22(1), 54-79. Wodon, Q., Onagoruwa, A., Malé, C., Montenegro, C., Nguyen, H. & De la Brière, B. (2019). How Large Is the Gender Dividend? Measuring Selected Impacts and Costs of Gender Inequality. The Cost of Gender Inequality Notes Series. Washington, DC: The World Bank. Wolfe, D. A., Crooks, C., Jaffe, P., Chiodo, D., et al. (2009). A school-based program to prevent adolescent dating violence: A cluster randomized trial. Archives of pediatrics & adolescent medicine, 163(8), 692-699. World Bank. (2020a). Africa’s Pulse, No. 21, Spring 2020 : An Analysis of Issues Shaping Africa’s Economic Future. Washington, DC: The World Bank. World Bank. (2020b). Poverty and Shared Prosperity 2020: Reversals of Fortune. Washington, DC: The World Bank. World Bank. (2020c). The COVID-19 Pandemic: Shocks to Education and Policy Responses. Washington, DC: The World Bank. World Bank. (2020d). Realizing the Future of Learning: From Learning Poverty to Learning for Everyone, Everywhere. Washington, DC: The World Bank. World Health Organization Global Consultation on Violence and Health. (1996). Violence: A Public Health Priority. Geneva: World Health Organization. World Health Organization. (2002). World Report on Violence and Health. Geneva: World Health Organization. World Health Organization. (2016). INSPIRE: Seven Strategies for Ending Violence Against Children. Geneva: World Health Organization. 81 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 World Health Organization. (2018). INSPIRE Handbook: Action for implementing the seven strategies for ending violence against children. Geneva: World Health Organization. World Health Organization. (2019). School-based Violence Prevention: A Practical Handbook. Geneva: World Health Organization. Yohalem, N., & Wilson-Ahlstrom, A. (2010). Inside the Black Box: Assessing and Improving Quality in Youth Programs. American Journal of Community Psychology 45, 350–357. Zingraff, M. T., Leiter, J., Johnsen, M., Myers, K. A. & Johnsen, M. C. (1994). The Mediating Effect of Good School Performance on the Maltreatment-Delinquency Relationship. Journal of Research in Crime and Delinquency, 31(1):62-91. Zolotor, A., Kotch, J., Dufort, V. et al. (2019). School performance in a longitudinal cohort of children at risk of maltreatment. Matern. Child Health J., 3(1):19-27. JULY 2021 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | 82 83 | ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE | JULY 2021 ENDING VIOLENCE IN SCHOOLS: AN INVESTMENT CASE QUENTIN WODON, CHLOË FÈVRE, CHATA MALÉ, ADA NAYIHOUBA, AND HOA NGUYEN JULY 2021