4.2.1 Bans against Child Marriage

Since the 1948 United Nations’ (UN) Universal Declaration of Human Rights the principle of granting equal marriage rights to men and women, respecting their free and full consent, has set the legal framework for banning child marriage and has been followed by subsequent international agreements. The 1962 UN Convention on Consent to Marriage, Minimum Age for Marriage, and Registration of Marriages and the 1979 UN Convention on the Elimination of All Forms of Discrimination Against Women require signatory nations to establish a legal minimum marriage age and ensure the registration of all marriages (United Nations, 1962; 1979). While the 1989 UN Convention on the Rights of the Child formally defines a child as an individual under the age of 18 (United Nations, 1989), it is the 1995 Beijing Declaration and Platform for Action that urged nations to enforce a legal minimum age for marriage, reaffirming 18 as the minimum age (United Nations, 1995). In 2016, the elimination of child marriage by 2030 became an international commitment under goal 5 on gender equality in the Sustainable Development Goals, resulting in a rapid increase in policy and programmatic efforts to end the practice.

In the past few years several countries introduced laws banning child marriage. While in 1995, more than 80% of countries allowed marriage for girls under 18 (with or without parental consent), by 2023 this rate had dropped to 33.16%. Yet, despite recent progress, according to the UN Children’s Fund (2023), overall progress would need to be 20 times faster to end child marriage by 2030. Figure 9 shows that worldwide only 8.3% of countries – mostly in the Middle East, North Africa, Sub-Saharan Africa, and South Asia – allow for child marriage without parental consent for a girl to marry under 18 years old. The requirement of parental consent creates, in fact, a legal loophole to child marriage as most early marriages are arranged with parental involvement, but this does not grant any further child protection relative to countries where no legal parental consent is required.

Figure 9 Share of countries allowing for child marriage, by macroregion.

Notes: Author’s calculation based on data from the World Policy Center (2023). Child marriage is considered allowed when the minimum legal age of marriage for girls is 17 or younger.

Recent studies assess the effectiveness of child marriage bans across countries. Batyra and Pesando (Reference Batyra and Pesando2021) adopt a cross-national perspective comparing six low- and middle-income countries (Benin, Bhutan, Kazakhstan, Mauritania, Nepal, and Tajikistan) that raised the minimum age at marriage over the past two decades. They find the reforms were unsuccessful in decreasing child marriage. According to the authors, the level of enforcement remains weak, and legal exceptions based on parental consent and customary or religious laws remain in place.

The study by Collin and Talbot (Reference Collin and Talbot2023) encompasses 90 developing countries between 1995 and 2012, out of which more than half did not have any legal ban against child marriage at the time of the study. The authors report a 60% prevalence rate of child marriage (legal and illegal) in 1980, which dropped by half in 2012. The main results, based on a cross-country and a within-country analysis, show poor law enforcement whereby legal bans were ineffective against child marriage. Adopting a regression discontinuity approach, the authors find that the age of marriage distribution does not show a discontinuity around the legal cutoff that we would expect if child marriage bans were effective. A weak law enforcement is found especially in rural areas.

Limiting the analysis only to 17 low-income countries that had already adopted a legal ban against child marriage, Wilson (Reference Wilson2022) implements an event-study approach to investigate changes in child marriage associated with the timing of the legal ban across areas with different intensity levels of child marriage before the ban.Footnote ¹⁶ Among these selected countries more than one-third of female respondents were married before 18 years old, and among ever married respondents, the average age of first marriage was nearly 19 years old. The results show a 20% significant decrease in child marriage only in urban areas, but not in the overall sample or in rural areas. Additional results suggest that child marriage bans seem more effective at decreasing child marriage in countries with a preexisting legal minimum age of marriage (16 or 17 years old). Looking at specific age of marriage, the author finds that bans are particularly effective at discouraging child marriage of girls aged between 13 and 16 years old.

Child marriage bans also appear to improve the girls’ lives in a variety of dimensions. Wilson (Reference Wilson2022) observes a delay in the age of first birth (one standard deviation increase in ban intensity is associated with an approximately 0.42 year increase in age at first birth), the effect being larger in urban areas. There are positive effects also on the girls’ number of years of education in urban areas only (+1 additional year) and on female labor force participation, both in urban and rural areas.

The study by McGavock (Reference McGavock2021) similarly reveals heterogeneous effects of banning child marriage. Looking at the laws implemented across Ethiopia’s semiautonomous regions, she finds a delay in women’s marriage, in particular for girls under 16 by about 17% (6.8 percentage points) in areas where early marriage was more common prior to the reform. However, the effect of the reform, though larger, is insignificant among women belonging to ethnic groups with the strongest norms toward early marriage.

Overall these results highlight that banning child marriage can be a feasible solution, although in the absence of adequate enforcement, the law’s impact remains limited and its effectiveness may vary across contexts. Among the various possible reasons for these mild effects is the presence of particularly tight social norms about age at marriage among some social groups, as the study by McGavock (Reference McGavock2021) documents. The weak law enforcement may also reflect the presence of powerful elites opposing change in favor of gender equality and undermining the implementation of progressive policies (Robinson & Acemoglu, Reference Robinson and Acemoglu2012; Kelsall et al., Reference Kelsall, Schulz, Ferguson, Vom Hau, Hickey and Levy2022). Exploring the reasons for this limited and heterogeneous effect seems a promising avenue for future research to assess the law’s enforcement capacity under different local constraints.

4.2.2 Cash and In-Kind Transfers Conditioned on Girls Staying Unmarried

As documented by Corno et al. (Reference Corno, Hildebrandt and Voena2020) and Corno and Voena (Reference Corno and Voena2023) child marriage is (also) a risk-coping strategy whereby economically constrained households resort to marrying off their daughters in response to negative income shocks, essentially in societies practicing brideprice. This mechanism highlights the binding role of the monetary constraint, which is possibly linked to the lack of insurance and credit access as predominant market failure. A possible policy solution to address these constraints would then be to provide cash transfers to poor eligible households for them to avoid resorting to child marriage in the presence of a negative income shock.

One possible way for cash transfers to delay child marriage is by attaching the conditionality requirement to the girls staying unmarried. Empirical evidence testing for the effectiveness of this type of conditional cash transfers shows, however, mixed results. The study by Erulkar, Medhin, and Weissman(Reference Erulkar, Medhin and Weissman2017) evaluates the Berhane Ewhan multi-arm program that included transferring livestock to households under the condition of 12–17-year-old girls remaining unmarried and in school. The program was implemented across sampled communities in Ethiopia and Tanzania.Footnote ¹⁷ In Ethiopia, the main results show a one-tenth decrease in child marriage among girls aged 10 to 14 years old – but no effect for those aged 15 to 19 years old. In turn, in Tanzania the girls aged 15 to 17 years old saw their risk of ever being married relative to the control group cut by half, though no effect was found on the 12–14-year-old girls.

The Apni Beti Apna Dhan (Our Daughter, Our Wealth) program implemented in the Indian state of Haryana consists in providing a financial grant transferred to households upon a daughter’s birth, followed by a long-term savings bond redeemable by the unmarried daughter at the age of 18. Sinha and Yoong (Reference Sinha and Yoong2009) document the program’s positive effects on the sex ratio and on girls’ health - but not on education. A related study by Nanda et al. (Reference Nanda, Das, Datta, Lamba and Pradhan2016) finds no significant decrease in child marriage, whereas they show that beneficiaries were more likely to marry at 18, probably using the cash transfer to fund their wedding.

A study by Buchmann et al. (Reference Buchmann, Field, Glennerster, Nazneen and Wang2023) evaluates the impact of a multi-arm program implemented by the nongovernmental organization Save the Children in Bangladesh over two years (2008–2010). The program consisted of two randomized interventions: life-skills training conducted for six months at the community level and targeted at girls aged 10–19 years old and a conditional asset transfer consisting in cooking oil transferred every four months for two years to households with a girl aged 15–17 years old under the condition of the girl remaining unmarried. In a subset of communities the two interventions were provided simultaneously. The main results show a 4.9 percentage point decrease (equivalent to 16.7% relative to the control mean) in child marriage achieved by the asset transfer alone. This leads to an increase in the age of marriage by 0.21 years, suggesting that girls marry soon after their 19th birthday (average marriage age in the control group is 18.969). In contrast, the life-skills training, either alone or combined with the asset transfer, has no significant impact on child marriage. The authors explain their results with a theoretical model whereby delaying marriage is a negative signal of unobservable bride characteristics that grooms desire, such as docility and obedience, pushing all girls to marry earlier than preferred to avoid negative signaling. This makes the authors exclude the hypothesis that child marriage is due to financial constraints, as they do not observe any premium attached to higher age of marriage – such as higher dowry or lower groom quality.

4.2.3 Life Skills, Vocational Training and Job-Market-Related Interventions

Similarly to the work by Buchmann et al. (Reference Buchmann, Field, Glennerster, Nazneen and Wang2023), several studies evaluate multi-arms interventions to test for the effectiveness of different approaches to fight child marriage.Footnote ¹⁸ Life-skills training is sometimes combined with vocational training or access to the labor market with the aim of tackling multiple pathways simultaneously. Life-skills training aims at avoiding child marriage by empowering the girls themselves, making them more capable of negotiating later marriages. This type of training is often coupled with community-level awareness campaigns to shift social norms. Meanwhile, improving the girls’ job skills and facilitating their access to the labor market aims at providing a viable alternative to marriage, while at the same time enhancing the girls’ financial autonomy, as household financial constraints are still often considered to drive child marriage decisions.

Work by Amin, Saha, and Ahmed (Reference Amin, Saha and Ahmed2018) evaluates the Bangladeshi Association for Life skills, Income, and Knowledge for Adolescents (BALIKA) program, whereby three different interventions targeted girls aged 12–18: an educational tutoring support, training for promoting gender rights awareness, and vocational training. In addition, all girls were offered a 44-hour life-skills training program over 18 months and an awareness campaign was conducted at the community level. Results show that all three interventions successfully reduce the risk of child marriage relative to the control group by about 25–30%.

A similar approach relying on training adolescent girls in “safe spaces” is adopted by the BRAC’s Empowerment and Livelihood for Adolescent Girls (ELA) initiative. The study conducted by Bandiera et al. (Reference Bandiera, Buehren, Burgess, Goldstein, Gulesci, Rasul and Sulaiman2020) in Uganda targeted girls aged 14–20 who were offered life-skills training combined with vocational training. Despite a low attendance rate (21%), the intervention managed to decrease the likelihood of being married by 6.8 percentage points, a 50% reduction relative to the control mean. However more mixed results are found by two other studies evaluating the same intervention in different contexts.Footnote ¹⁹ No significant impacts are found on marriage and childbearing in Tanzania (Buehren et al., Reference Buehren, Goldstein, Gulesci, Sulaiman and Yam2017) while an increase in marriage and childbearing rates is found in South Sudan (Buehren et al., Reference Buehren, Chakravarty, Goldstein, Slavchevska and Sulaiman2017). Among the possible reasons discussed by the authors to explain these divergent findings stems the different context, different implementation teams, interventions implemented at different levels of scale, with different amounts of resources available. Besides, in the South Sudan intervention eligible girls were slightly older (15–24 years old). Finally, Boulhane et al. (Reference Boulhane, Boxho, Kanga, Koussoubé and Rouanet2024) tested for the effectiveness of training girls aged 8–24 years old on life skills in safe spaces implemented in Ivory Coast. While the stand-alone intervention had limited impacts on girls’ empowerment, it had the largest impacts when coupled with similar safe spaces for boys and men of the same age group. Despite encouraging effects on girls’ sexual and reproductive health, economic activities, and decision-making, the study does not find any significant decrease in girls’ marriage or pregnancy.

To discourage early marriage, another approach is to facilitate girls’ direct access to the labor market. The intervention studied by Jensen (Reference Jensen2012) tackles girls’ access to the labor market by conducting information campaigns and recruitment initiatives in rural villages in the outskirts of Delhi. The intervention randomly assigned villages to be connected to recruiting services of business process outsourcing (BPO) industry – such as call centers and back-office services. The BPO industry represents a new fast-growing economic sector that employs mostly female workers with at least 10–12 years of schooling. The study finds that girls aged 18–24 years, mostly unmarried, are 4.6 percentage points more likely to be working for a BPO industry one year later. Interestingly, this impacts their family outcomes. Even though the age composition of the sample does not allow researchers to explore effects on child marriage, results show a decrease in the likelihood of being married and of childbearing by about 5 percentage points. The intervention has also positive spillovers on younger girls aged 6–17 who are 5 percentage points more likely to be enrolled in school, closing the educational gender gap relative to boys.

Similarly to the BPO industry, in Bangladesh the ready-made garment industry has spiked since the 1990s, accounting today for 75% of the country’s export earnings. This sector offers a large-scale employment opportunity for women with basic literacy and numeracy. The study by Heath and Mobarak (Reference Heath and Mobarak2015) assesses the impact of the sector’s expansion on women’s family life by relying on retrospective data from households living in 60 Bangladeshi villages that vary with respect to their distance to garment factories, and to the date when the first factories opened close to each village. The authors show that girls living in villages close to a garment factory when aged between 10 and 23 years old – that is the window that covers most of the marriage age in the sample – are 13 percentage points more likely to work outside of the home and are 0.3 percentage points less likely to be married for the average number of years of exposure to the garment factory (6.4 years), corresponding to a decrease of 28% relative to the control mean. The effect sizes are the largest for girls aged 12–18. The likelihood of childbearing decreases too, by 0.23 percentage points, corresponding to 29% relative to the control mean.

The empirical evidence available to date seems to point toward girls’ access to the labor market as the most effective policy against child marriage, with impact sizes similar to those of passing legal bans found by Wilson (Reference Wilson2022). Evidence about life-skills and vocational training shows, in turn, more mixed results, making it hard to conclude whether this type of intervention is successful against child marriage. Divergent findings suggest that interventions of this type may be highly sensitive to context, implementation quality, and available resources. Moreover, girls are targeted at a relatively young age, most often before their 15th birthday, making parental permission and acceptability key for ensuring high participation rates. Provided the strong social norms about age at marriage, targeted training interventions may benefit from complementary community campaigns to grant a wider endorsement.

4.2.4 Education-Related Interventions

Another approach to tackle child marriage relies on education programs. In a similar spirit as providing job skills or facilitating access to the labor market, schooling programs aim at increasing girls’ human capital or opportunities by making schooling a viable alternative to marriage. Schooling and marriage are often seen as mutually exclusive, such that for as long as girls stay in school their marriage is delayed (incarceration effect). Moreover, education might reinforce the girls’ skills, confidence, and aspirations, making them able to negotiate a later marriage. Reaching a critical mass of girls enrolled in school – especially at the lower-secondary level – might also encourage a general shift in social norms by pushing parents to delay the marriage decision.

A study by Baird, McIntosh, and Özler (Reference Baird, McIntosh and Özler2011) evaluates the effect of providing a conditional cash transfer (CCT) versus an unconditional cash transfer (UCT) to girls aged 13–22 in rural villages of Malawi. The CCT was conditioned on the girls’ school attendance, targeting girls already in school or who dropped out from school. In turn, the UCT provided cash without any condition and targeted only girls enrolled in school. Transfers took place over two years. At the end of the program, the CCT was found to be effective in increasing girls’ school attendance and test scores. But it was only the UCT that was effective at decreasing the likelihood of marriage (by 8 percentage points, equivalent to a 44% decrease relative to the control mean) and childbearing (by 6.7 percentage points, corresponding to a 27% decrease relative to the control group).

This shows that while CCTs may have an indirect effect on marriage and childbearing through increased school participation, UCTs can have a direct effect through income. Conditionality was in this case effective for education-related outcomes, but it undermined social protection by denying support to households who could not comply (anymore) with the conditionality requirement. Nevertheless, these positive effects brought by the UCT were not sustained two years after the end of the program. As Baird, McIntosh, and Özler (Reference Baird, McIntosh and Özler2019) show, marriages and childbearing spike among UCT beneficiaries soon after the end of the program, leading to lower levels of empowerment compared with both the CCT and the control groups. In contrast, girls who were out of school at baseline and who benefited from the CCT reported lower marriage and pregnancy rates two years after the end of the program.

The lack of sustained effect of antipoverty UCT programs is also confirmed by Dake et al. (Reference Dake, Natali and Angeles2018) who find no significant effects of a UCT implemented in Malawi and Zambia on early marriage of girls aged 14–21 years old. In turn, the capacity of CCTs and school subsidies to decrease early marriage and pregnancy by keeping girls in school is confirmed by other studies in several different contexts (Angrist et al., Reference Angrist, Bettinger, Bloom, King and Kremer2002; Duflo, Dupas, & Kremer, Reference Duflo, Dupas and Kremer2015; Amin et al., Reference Amin, Saha and Ahmed2018; Hahn et al., Reference Hahn, Islam, Nuzhat, Smyth and Yang2018; Giacobino et al., Reference Giacobino, Huillery, Michel and Sage2024). This literature shows that, by keeping girls longer in school, marriage is delayed. The evidence showing long-lasting effects (see, for instance, Hahn et al., Reference Hahn, Islam, Nuzhat, Smyth and Yang2018) provides support to the hypothesis that education acts on the age of marriage not only through an incarceration effect – lasting for as long as the girl is in school but dissipating soon after – but as a transformative process that enhances girls’ empowerment. An open question remains about the pivotal mechanisms that make CCTs successful at delaying marriage: whether they mostly act on the girls’ side, by improving their human capital, financial autonomy, and decision-making power; whether they successfully shift social norms about age at marriage by reaching a critical mass of households; whether they act on the parents’ side by increasing their perceived returns to girls’ education; or whether they trigger some other type of mechanisms, or even a combination of all these. Currently, different studies point toward different mechanisms, as these may, in fact, vary with context and intervention design.

It is worth mentioning that the existing literature on educational programs addressing girls’ outcomes, and more specifically child marriage, is limited to the demand side of education programs. That is, the focus is on providing households with cash transfers or other support programs for encouraging girls’ enrollment and attendance. A yet unexplored but promising avenue of research is the analysis of supply-side interventions – for example, school building, improving school infrastructures, female teachers, content of curriculum, etcetera – on girls’ outcomes.

Overall, school subsidies and conditional cash or in-kind transfers appear to be the most effective method for delaying marriage, more successful than family economic support, such as antipoverty UCTs. Facilitating access to the labor market also appears successful in delaying marriage and childbearing. The study by Jensen (Reference Jensen2012) shows that providing household with concrete opportunities for their daughters’ employability effectively shifts attitudes and behaviors, leading to intergenerational effects. In turn, enhancing girls’ human capital, particularly through life skills, livelihoods, and gender rights training, shows more mixed results.

Besides their effectiveness, some of the interventions enhancing girls’ human capital and work opportunities appear to be scalable. The studies by Jensen (Reference Jensen2012) and by Heath and Mobarak (Reference Heath and Mobarak2015) shed light on the effects of women’s employment in fast-growing economic sectors that operate nationwide, possibly impacting millions of women. While few studies report information about their cost-effectiveness and capacity to operate at scale, the Colombian government school voucher program (Angrist et al., Reference Angrist, Bettinger, Bloom, King and Kremer2002; Angrist, Bettinger, & Kremer, Reference Angrist, Bettinger and Kremer2006) and the Bangladeshi government program for school support (Hahn et al., Reference Hahn, Islam, Nuzhat, Smyth and Yang2018) are examples of large-scale interventions that successfully reach vast numbers of girls and appear to be sustainable.

5.1.1 List Experiments

Among the experimental survey techniques used in the current literature to avoid the risk of misreporting, list experiments (LEs) have been widely adopted. Instead of directly asking respondents, LEs present them with a list of items and ask them the number of items they agree with. This survey technique is used in experimental surveys where respondents are randomly allocated to a control and a treatment group. In the control group, respondents are asked about a list of J  “baseline” items, while in the treatment group, respondents are asked about a list of J + 1 items, with the additional item being the sensitive one related to the research topic. Since respondents are randomly assigned to the two groups, the prevalence rates of the baseline items should be the same across both groups. Therefore, the difference in the average number of items reported in the treatment group versus the control group serves as an estimate of the prevalence of the sensitive item in the sample.

In addition, in the control group respondents are asked about the sensitive item in a standard direct way. Two measures of the sensitive item prevalence rate can be computed: one from the LE, based on the comparison between the counts declared by respondents in the two groups; and one from the direct question administrated to the control group. Any significant difference between the two prevalence rates should signal the presence of misreporting.

The use of an LE has the main advantage of avoiding any public disclosure of sensitive personal information, granting a higher confidentiality than standard direct questions and strongly limiting the risk of social desirability bias. The example reported in Table 2, taken from Bertelli et al. (Reference Bertelli, Calvo, Lavallée, Mercier and Mesplé-Somps2025), shows the structure of a typical LE compared to a standard direct question.

With regard to the practice of GBV, the existing literature shows that GBV victimization is often misreported – mostly underreported – by victims when asked standard direct questions (Plummer et al., Reference Plummer, Wight and Ross2004; Phillips et al., Reference Phillips, Gomez, Boily and Garnett2010; Joseph et al., Reference Joseph, Javaid, Andres, Chellaraj, Solotaroff and Rajan2017; Bulte & Lensink, Reference Bulte and Lensink2019; Traunmüller, Kijewski, & Freitag Reference Traunmüller, Kijewski and Freitag2019; Lépine, Treibich, & d’Exelle, Reference Lépine, Treibich and d’Exelle2020; Cullen, Reference Cullen2022). The size of the response bias is usually around 10 percentage points. Misreporting often appears to vary according to respondent characteristics, education standing out as the most common heterogeneity dimension. Under-reporting of IPV is usually higher among educated women (Joseph et al., Reference Joseph, Javaid, Andres, Chellaraj, Solotaroff and Rajan2017; Agüero and Frisancho, Reference Agüero and Frisancho2022; Cullen, Reference Cullen2022).Footnote ²⁰

Few studies have so far explored misreporting in the justification of IPV and in the attitudes toward GBV with experimental surveys. Surveying adolescent girls in Bangladesh, Asadullah et al. (Reference Asadullah, De Cao, Khatoon and Siddique2021) find that support for IPV and for child marriage is considerably underreported when directly asked. Support for IPV is also underreported by men in rural central Ethiopia (Gibson et al., Reference Gibson, Gurmu, Cobo, Rueda and Scott2022) and by men in Bamako (Bertelli et al., Reference Bertelli, Calvo, Lavallée, Mercier and Mesplé-Somps2025). Recent studies also document considerable response bias toward FGM, a form of GBV still present in Sub-Saharan Africa. De Cao and Lutz (Reference De Cao and Lutz2018) and Gibson et al. (Reference Gibson, Gurmu, Cobo, Rueda and Scott2018) show that support for FGM is underreported by men and women in Ethiopia, where FGM is banned. In contrast, in Mali, where FGM is not illegal, Bertelli et al. (Reference Bertelli, Calvo, Lavallée, Mercier and Mesplé-Somps2025) show that support for FGM is overreported. This study also shows that in Mali, where the minimum legal age for girls to marry is 16, men do not underreport their support for child marriage (that is, under 18 years old), contrary to the findings of Asadullah et al. (Reference Asadullah, De Cao, Khatoon and Siddique2021) where the practice is illegal.

6.1.1 Legislation

As documented in Section 2, worldwide 40% of women and girls live in countries where laws and social institutions still exhibit a high or very high level of discrimination against women (OECD, 2023). These formal legal rules often mirror and perpetuate discriminatory social norms. Unequal inheritance laws, for example, entrench women’s economic dependence by limiting their access to property and intergenerational wealth. Likewise, laws setting different minimum marriage ages for girls and boys not only reflect existing customs but also reinforce the idea that early marriage for girls is acceptable. Such legal disparities shape gendered attitudes, fostering the belief that women should remain financially reliant on a father or husband and play a lesser role in economic life. By embedding these inequalities into law, legal systems help sustain gender gaps in labor market participation and decision-making power.

In principle, introducing laws, with associated sanctions, against discriminatory practices could successfully curb those behaviors ruled by social norms. The sanctions would alter the cost and benefit of the targeted behavior, changing empirical and normative expectations. In theory, people are predicted to abide by the law if sanctions are sufficiently severe whereas they would tend to break the law if sanctions are too mild. However, to grant successful enforcement the law and the authorities should also be seen as legitimate and, probably even more importantly, the legal arrangements should not appear so distant from existing social norms as to lose credibility. The challenge for legal reforms aimed at changing social norms is that they should promote new behaviors while approximating popular views, otherwise the threat to seek enforcement will not be credible (Bicchieri & Mercier, Reference Bicchieri and Mercier2014).

The weight of social norms is such that it can potentially hamper the effectiveness of legal reforms and targeted interventions. As illustrated at various occasions across this Element, the scientific evidence on the effectiveness of policy reforms in favor of gender equality and equity shows mixed results. For instance, as discussed in Section 3.1, legislation imposing equal inheritance rights for men and women has proven ineffective amid strong patrilineal social norms in the Indian context. Similarly, a comparative review of land rights reforms in Kenya, Rwanda, and Uganda documents that granting equal rights to land while promoting land titling and privatization is insufficient – and even detrimental – due to the prevailing patriarchal customary laws and patrilineal social norms that prevent women from owning any land (Djurfeldt, Reference Djurfeldt2020). The implementation of gender equity policies may also suffer depending on the structures and dynamics of powerful groups, whether their interests align with those of the legal reforms or whether they can successfully act against them (Nazneen, Hickey, & Sifaki, Reference Nazneen, Hickey and Sifaki2019).

Legal reforms have also addressed GBV to punish perpetrators, in spite of a persistent and widespread justification of IPV (see Section 5). Worldwide, most countries have passed legislation against GBV, though with varying degrees of definitions and sanctions. As of 2023, out of 189 countries, only 25 report no legislation against GBV, mostly in the MENA region.

Recent evidence shows that laws against domestic violence can, in fact, change behaviors and social norms. Sanin (Reference Sanin2024) studies the 2008 Rwandan domestic violence law and the 2010 Ugandan law, which was followed in 2015 by a Supreme Court decision that made it unconstitutional to refund the brideprice upon divorce. The Rwandan anti-GBV legislation is one of the most comprehensive and transformative ones that directly challenges strong patriarchal norms, not only by banning domestic violence, but also by outlawing nonconsensual marital sex defined as rape. Born out of the ashes of the 1994 conflict, the women’s groups advocacy work was strongly backed up by key members of the parliament and government actors, which also granted effective implementation (Burnet, Reference Burnet, Nazneen, Hickey and Sifaki2019). As the study by Sanin (Reference Sanin2024) shows, in Rwandan genocide-intense areas – where male scarcity led women to marry more violent men – divorce rates went up and domestic violence decreased after the law enactment. Interestingly, the results point toward an intergenerational transmission of different gender norms. The younger generation living in genocide-intense areas exhibits higher bargaining power, a reduction in domestic violence, and its acceptance without a rise in divorce rates after the law enactment.

These results suggest that laws can act as catalysts for changing social norms and behaviors, but, likely, under certain circumstances. In particular, assuring wide societal and governmental support, whereby key powerful groups and elites back up the reform, seems key. As put forward by Acemoglu and Jackson (Reference Acemoglu and Jackson2017), the gradual tightening of laws is effective when in line with prevailing social norms. In turn, when laws are in conflict with norms, tighter laws can have counteracting effects, reducing behavior among law-abiding individuals but increasing it among lawbreakers. In line with this, Sanin (Reference Sanin2024) shows that the anti-GBV legislation passed in Uganda was not effective, possibly because it suffered from a lack of support from key powerful groups, which might be symptomatic of the law being too much in conflict with mainstream social norms about domestic violence.Footnote ²⁴ In contrast, changing the cost of marital dissolution appeared to be more effective at curbing domestic violence. After the Supreme Court decision reduced the cost of divorce by prohibiting the refund of the brideprice, and dissolution was already socially acceptable, women from ethnic groups where the brideprice was previously refundable were more likely to separate and experienced less domestic violence (Sanin, Reference Sanin2024). The importance for legislation to be effective to count on societal support and/or on support of powerful groups is confirmed by Gulesci, Leone, and Zafar (Reference Gulesci, Leone and Zafar2024) who document the risk of laws backfiring. The study shows that Pakistan’s new domestic violence laws increased women’s self-reported IPV in ethnic groups with very conservative views on wife-beating and divorce. The authors argue that this illustrates the backlash effect of laws that are in stark conflict with prevailing social norms.

Another dimension of gender inequality on which an increasing number of countries has passed legislation are traditional harmful practices such as child marriage and FGM. The evidence here points toward a limited capacity of laws to curb these practices. In the case of child marriage, studies generally find limited or no impact of legal bans, probably due to weak law enforcement and persistent social norms (Batyra & Pesando, Reference Batyra and Pesando2021; McGavock, Reference McGavock2021; Collin & Talbot, Reference Collin and Talbot2023; Wilson, Reference Wilson2022).Footnote ²⁵ With respect to FGM, while some evidence suggests that legal bans can reduce FGM rates (Crisman et al., Reference Crisman, Dykstra, Kenny and O’Donnell2016; García-Hombrados & Salgado, Reference García-Hombrados and Salgado2023), other studies highlight significant limitations (Camilotti, Reference Camilotti2015; Cetorelli et al., Reference Cetorelli, Wilson, Batyra and Coast2020; Kudo, Reference Kudo2023).

By comparing Mauritania, which banned FGM in 2005, with Mali, where no such law exists, Cetorelli et al. (Reference Cetorelli, Wilson, Batyra and Coast2020) found no significant difference in the prevalence of FGM. In contrast, the introduction of an anti-FGM law in 1996 in Burkina Faso resulted in a significant decline in the prevalence of the practice, possibly thanks to the interplay between legislation and robust grassroots campaigns, which fostered societal disapproval of the practice and increased the perceived costs of compliance with traditional norms (Crisman et al., Reference Crisman, Dykstra, Kenny and O’Donnell2016). However, Kudo (Reference Kudo2023) reports that the reduction was less pronounced for severe forms of the practice involving younger girls, underscoring the resilience of cultural practices that are perceived as important to community identity. The 1999 ban enacted in Senegal was found to effectively reduce the likelihood of undergoing FGM and simultaneously increased educational investments in girls, as parents reallocated resources in response to the heightened costs associated with FGM (García-Hombrados & Salgado, Reference García-Hombrados and Salgado2023). This contrasts with previous findings from Camilotti (Reference Camilotti2015), who documents that anti-FGM laws in Senegal prompted families to perform the practice earlier in a child’s life to avoid detection, resulting in increased health risks due to the clandestine nature of the procedures.

It is somewhat hard to reconcile these divergent findings, as the effectiveness of a law is likely highly context-specific, depending on the country’s political and social dynamics. A study by Poyker (Reference Poyker2023) shows, indeed, that anti-FGM laws tend to be more effective in politically stable regimes, where people have greater trust in the consistency and durability of government enforcement. In contrast, in politically unstable settings, uncertainty about the enforcement of laws can weaken efforts to change deeply rooted social norms. This likely influences the capacity of legal bans to successfully change normative social norms by establishing a credible change in sanctions, in that the original negative social sanction for not following the norm will be substituted by a new, credible negative sanction for following it. In this case, normative expectations would change, too (Bicchieri & Mercier, Reference Bicchieri and Mercier2014).

6.1.2 Media and Edutainment

Beyond formal legislation, efforts to shift gender norms have included nationwide mass media campaigns or campaigns targeted at the community level. Exposure to information and different role models broadcast by the mass media has proven especially influential in shaping attitudes and behaviors. Despite not having this explicit goal, radio and television serial dramas that model gender-equitable attitudes, for example, portraying a father supporting his daughter’s education, or a husband respecting his wife’s decisions, have also been associated with shifts in audience attitudes and behaviors.

In Brazil, access to soap operas portraying small families led to lower fertility among women (Ferrara, Chong, & Duryea, Reference Ferrara, Chong and Duryea2012) and to increased divorce and separation rates (Chong & Ferrara, Reference Chong and Ferrara2009). In India, access to cable TV is associated with improved gender attitudes (decrease of son preference and of domestic violence justification) and women’s behavioral outcomes (increased autonomy, lower fertility) (Jensen & Oster, Reference Jensen and Oster2009). In Egypt, an anti-FGM campaign broadcast on radio between 1994 and 2003 reduced the likelihood of being cut by 13% among exposed cohorts under 15 years old at the time of the campaign (Khalifa, Reference Khalifa2022). However, there is also evidence of backlash effects. Exposure to telenovelas with LGBTQ+ characters lowered tolerance toward homosexuals and homosexuality across 14 Latin American countries, the effect being larger among more conservative individuals (Gulesci, Lombardi, & Ramos, Reference Gulesci, Lombardi and Ramos2024).

Beyond exposure to mass media, the evidence about the capacity of targeted information campaigns or video messages to change gender-related behaviors and social norms is more mixed.Footnote ²⁶ A recent review by Reference PetermanPeterman (2025) looking at edutainment interventions on violence against women and children reports that most interventions are often found to effectively curb attitudes in support for violence, at least in the short term. However, their capacity to change behaviors is limited as most studies report null or small effects on violence against women.

In her review, Peterman (Reference Peterman2025) reports that edutainment interventions against child marriage (eight studies) seem slightly more successful at curbing actual behaviors.Footnote ²⁷ Among these, four measure impacts on child marriage rates and three find a significant decrease. A study by Raghunathan, Sushant, and Mankad (Reference Raghunathan, Sushant and Mankad2021) finds a significant decrease in child marriage (− 9 percentage points) and a delay in average age at marriage (+6.5 months) still five years after the end of the Breakthrough edutainment program - a large-scale intervention implemented in two Indian states combining a mass-media campaign urging men to act against child marriage, training about gender rights, sexuality, and sexual harassment targeted at key stakeholders at the community and Grand Panchayat level, and school-based training against early marriage. Interestingly, the delay in child marriage is coming from keeping girls longer in school (+9 months). Positive effects are also found in north Nigeria where a targeted edutainment intervention that successfully improved boys’ and girls’ primary school attendance and learning was also found to decrease the combined likelihood of teenage childbearing and marriage by 1.4–3.9 percentage points 12 months after the intervention (Orozco-Olvera & Rascon-Ramirez, Reference Orozco-Olvera and Rascon-Ramirez2023). These results combined echo those presented in Section 4.2.4 whereby several interventions find that increasing (girls’) schooling manages to delay age at marriage. Moreover, these studies add an interesting piece of evidence: acting on the side of preferences with edutainment interventions is an additional successful way for promoting schooling and deterring child marriage.

Interesting insights about who shall be targeted within a community to transform gender social norms are provided by Cassidy et al. (Reference Cassidy, Dam, Janssens, Kiani and Morsink2024). The study evaluates whether it is more successful to target an edutainment intervention – a mobile cinema screening of a street-theater performance against child marriage in two provinces of Pakistan – to men, women or both. Targeting men only significantly reduces the probability of female child marriage six months (− 4.1 percentage point decrease, a 66% reduction relative to the control mean) and one year and a half after the end of the intervention (− 5.3 percentage point decrease, a 43% reduction relative to the control mean). Targeting women only, who are considered more hesitant to depart from social norms about child marriage, is not effective, neither in the short nor in the long run. Targeting both women and men within the same household is effective only in the long run (− 5.2 percentage points), but not in the short run. The authors reconcile these results with a theoretical model whereby women attach greater weight to the sanctions of deviating from community norms than men do. Consequently, men would be more receptive of the intervention and would be more successful at persuading their untreated spouses to delay their daughters’ marriage. However, they would be somewhat less successful in households where both are treated. To be convinced, mothers would first need to perceive the shift in community norms, making them accept to delay their daughters’ marriage only in the long run. Women are found, notwithstanding, an important actor of change at the community level. Targeting women – either alone or jointly with their husbands – translates into similarly lower child marriage rates at the village level. According to the authors, mothers – given their stronger reluctance to deviate from social norms – would have greater incentives than fathers to diffuse information within the community to shift social norms, and they would do so through gendered networks.

These studies provide some new interesting insights about the mechanisms surrounding changes in social norms. Raghunathan et al. (Reference Raghunathan, Sushant and Mankad2021) highlight that, despite a significant decrease in child marriage rates, the intervention did not make households hold more progressive norms about gender roles within the home. In fact, education is perceived as a vehicle for better management of the home rather than for empowering girls. Therefore, systematically interpreting changes in behaviors as changes in social norms can be misleading. Furthermore, behavioral changes at the household or at the community level may follow very distinct dynamics, especially when people strictly abide to community normative norms, as documented by Cassidy et al. (Reference Cassidy, Dam, Janssens, Kiani and Morsink2024).

Several mechanisms seem to be at play. First, mass-media campaign and targeted edutainment interventions may improve people’s knowledge, for instance, about the adverse consequences of violence, child marriage, or FGM.Footnote ²⁸ However, behaviors dictated by normative social norms are hardly due to a lack of information, but most often to the fear of deviant behaviors being sanctioned by the reference group. Hence, the knowledge channel may, at best, affect individuals’ preferences, but only change behaviors when these are not governed by normative social norms. Second, these interventions can also act on individual persuasion, whereby people are exposed to role models who behave in a different way or who directly tell them that their normative beliefs are mistaken.Footnote ²⁹ This would be relevant for updating personal beliefs and changing behaviors dictated by descriptive norms, conventions, and habits. However, it might not be enough to change normative social norms as people need to be convinced, in addition, that the large majority of their reference group is changing behavior and that deviating from the previous norm will not be sanctioned. Third, another possible channel is norms’ perceptions – most often measured as changes in second-order beliefs, meaning people’s beliefs about what the others think.Footnote ³⁰ Yet the fact that a shift in perceptions does not translate into a behavioral change in most of these studies suggests that most interventions may have only affected, at best, empirical expectations, but not normative ones. For instance, learning about the illegality of child marriage might not change behaviors if the cost of deviating from the social norm is higher than the cost of deviating from the law, or if the law is perceived as not legitimate, too distant from existing norms or weakly enforced. Similarly, changing the perception of the community support for FGM will translate into a behavioral change only if people also start to believe that its abandonment will not be followed by negative sanctions.

In light of these mixed results, partly linked to the wide diversity of the study designs, and to the empirical challenging measurement of normative social norms, the literature on edutainment interventions has still much to explore. Future research would be particularly welcome for examining whether this type of interventions effectively changes not only attitudes and second-order beliefs but also normative expectations in order to trigger a change in social norms and associated behaviors. More studies are needed to test theories on how role models and storytelling influence attitudes, norms, and actual behaviors.

6.1.3 Targeting Boys and Men

Recent efforts have tried to shift gender-related behaviors and social norms by focusing on men. These interventions aim to address the role of men in violence perpetration, recognizing the key role that masculinity and gender-related social norms play in the acceptability and use of violence against women. Although not all men are violent, and some actively oppose violence, the use of violence over women is one source of power accorded to men in many settings. Mainstream masculinity indeed often entails dominance and control over women, which are attributes usually constituting the most legitimate and acclaimed version of hegemonic masculinity (Connell, Reference Connell2013). Importantly, this dominant ideal of masculinity is not imposed but most often widely accepted as normal by both men and women from a young age. Most men aspire to it, even if they can’t fully achieve it (for instance, due to poverty preventing a man from being a provider), and it shapes shared views on how men and gender roles should be (Jewkes, Flood, & Lang, Reference Jewkes, Flood and Lang2015). Thus both men and women need to be actively engaged to change this predominant masculinity ideal and gender-related norms.

A wide variety of interventions involving men and aimed at violence prevention are implemented worldwide; however, rigorous evaluations are scarce. A study by Ricardo, Eads, and Barker (Reference Ricardo, Eads and Barker2011) reviews 65 high-quality studies implemented in 11 countries, out of which 4 took place in middle-income countries (Brazil, India, Korea, and South Africa) and 3 took place in low-income countries (Ethiopia, Nicaragua, and Thailand).Footnote ³¹ Overall, only 8 interventions met criteria that placed them in the category of “strongest” quality, while 21 were classified as of “moderate” quality.Footnote ³² Out of these 29 studies, 15 evaluated the effects of interventions on perpetration of sexual, nonsexual, or both types of violence against women, as self-reported by men. The effects on self-reported perpetration of violence seems rather limited. Only 1 study found a significant decrease in sexual violence and 7 reported a decrease in any form of violence against women.Footnote ³³ The reviewed interventions were more successful at shifting attitude towards violence (10 out of 16 ) and gender roles (5 out of 12), however for most studies there is a serious risk of these results being affected by social desirability bias as all outcomes were directly self-reported by men and little was done to address this risk of bias.

Recent evidence about engaging men alone in peer-facilitated group discussions aimed at transforming inequitable gender attitudes is provided by Vaillant et al. (Reference Vaillant, Koussoubé, Roth, Pierotti, Hossain and Falb2020) who evaluate the “Engaging Men through Accountable Practice” intervention with a clustered randomized control trial in North and South Kivu. The intervention consisted in a 16-week group-based discussion series for men with three-hour-long sessions led by male trainers. About six months after the end of the intervention, men were significantly less likely to intend to commit violence and to justify wife-beating, whereas they appeared more likely to accept women refusing sex for any reason. However, no significant reduction in women’s reports of physical or sexual IPV is found, even though women reported improvements in relationship quality and reductions in negative male behaviors.

Larger impacts were found by a closely related study investigating a randomized controlled trial intervention targeting couples in Rwanda (Doyle et al.,Reference Doyle, Levtov and Barker2018). Male participants were involved together with their female partners in a 15-session curriculum of 45 hours in total to discuss topics such as gender equality, fatherhood, reproductive health, and violence prevention.Footnote ³⁴ The study found that, two years after the intervention, men in the treatment group were significantly more involved in caregiving and household tasks, showed greater support for shared decision-making, and were less likely to perpetrate physical or sexual IPV. Importantly, their female partners also reported improved relationship quality and reduced experiences of violence. These findings underscore the potential of structured, couples-based programs to shift harmful gender norms and promote healthier, more equitable relationships.

A different approach is the one evaluated by Nguyen and Tarp (Reference Nguyen and Tarp2022), who tested two alternative types of interventions to encourage men to critically reflect about gender equality in Vietnam. The authors randomly assigned participants to two different tasks: either comment on national gender laws or write stories about gender equality. Those who wrote stories showed a 0.29 standard-deviation improvement in self-reported gender attitudes and a modest increase in housework contributions, while mere commentary had no effect. Simply pushing men to think about gender issues appears to change their perceptions of gender equality, even though it was not successful at changing their behavior, such as engaging in housework and providing childcare. Interestingly, this study also documents that providing information about gender-related laws does not change men’s perceptions of gender issues, a possibly informative takeaway for those studies investigating the effectiveness of laws against GBV.

The capacity of a large-scale intervention to improve men’s gender equitable attitudes and behavior is investigated by Alderman et al. (Reference Alderman, Gilligan, Hidrobo, Leight, Mulford and Tambet2025) with respect to Ethiopia’s graduation program Strengthen PSNP Institutions and Resilience. On top of livelihoods and nutrition activities, the program had an explicit aim of promoting gender equity in two ways. First, men and women were brought together in village economic and social associations to facilitate discussions around financial literacy, savings and credit, income-generating activities, and gender equitable norms. The gender-related topics were discussed over six one-hour bimonthly sessions.Footnote ³⁵ Second, men were invited to eight sessions of men-only group discussions to critically reflect on traditional gender roles and explore the positive and perceived negative effects of male involvement in tasks traditionally assigned to women.

Results show that at one-year follow-up men’s gender equitable attitudes and their involvement in household tasks significantly increase. However, by three-year follow-up, only treatment arms with the men’s groups continue to have significant impacts. Interestingly, the impact size is larger for men’s engagement in household tasks (0.26–0.29 standard deviations) – such as cleaning, cooking, collecting firewood and water – than for gender equitable attitudes (0.07–0.14 standard deviations). Similar results are found from women’s reports of their partner’s behavior, mitigating worries of social desirability bias. This study underscores the transformative capacity of peer groups where men encourage and support each other to redefine traditional gender roles, effectively shifting attitudes and behaviors.

In this regard, a study by Pulerwitz et al. (Reference Pulerwitz, Hughes, Mehta, Kidanu, Verani and Tewolde2015) confirms the importance of group dynamics for triggering changes in attitudes. The authors investigate the Male Norms Initiative, a community-based intervention aimed at shifting gender norms and reducing IPV perpetration among men aged 15 to 24 in Addis Abeba, Ethiopia. Using a quasi-experimental design – caution should be used in interpreting these findings due to small sample size and lack of proper randomization design – the community intervention was either implemented alone or in combination with at set of eight group training sessions delivered in youth centers. The community intervention alone or combined with group training effectively decreases self-reported perpetration of IPV among participants. However, it successfully improved participants’ support for gender-equitable norms only when combined with the group training. These results suggest that group discussion and education are an important component for interventions to effectively shift attitudes toward gender norms and masculinity.

While most studies focus on adult men, Dhar, Jain, and Jayachandran (Reference Dhar, Jain and Jayachandran2022) investigate adolescents’ gender-related attitudes. The study evaluates an intervention conducted in the Indian state of Haryana that lasted much longer than those previously presented here. It engaged 7th to 10th graders in classroom discussions about gender equality, with a 45-minute session held every three weeks for two and one-half school years.Footnote ³⁶ The program was found to significantly improve gender attitudes by +0.18 standard deviations in the short run (three and one-half months after the program ended) and by +0.16 standard deviations two years after the program ended. The program also appeared to improve self-reported gender-equal behavior, in both the short and medium run; however, more objective measures of behavior – whether girls submitted a college scholarship application and signing a public petition to end the dowry system – do not appear to improve. Interestingly, the study finds a significant and positive effect on attitudes and self-reported behaviors for both boys and girls in the short and medium run. However, boys report a larger and sustained impact on behaviors, whereas the effect on attitudes is less sustained for girls in the medium run. While the authors show that the results in the overall sample are not driven by experimenter demand effect, as measured by the Marlowe–Crowne social desirability scale, they do not show whether desirability bias matters differentially for boys and girls.

Even though the available scientific evidence is still probably too thin to clearly identify the key elements that make interventions targeting men and boys successful at changing gender social norms, a few patterns emerge. First, attitudes seem easier to change than actual behaviors, especially when it comes to perpetration of IPV. Part of the reason might be that changing attitudes does not equate with changing normative social norms (or their perception), making the last mile of a behavioral change hard to reach. Moreover, more evidence is needed to understand the type of men who effectively shift attitudes. Awareness might be more easily raised among those more sympathetic with gender equality and with the least propensity to ever be violent. In this respect, the selection of participants should provide enough variation among men to carefully avoid reinforcing gender-inequitable masculine ideals with the risk of further marginalizing men with different attitudes. Second, engaging the reference network, by reaching out to communities and not just individuals, seems to be more effective at addressing hegemonic gender norms than just focusing on individual-level attitude changes (Pulerwitz et al., Reference Pulerwitz, Hughes, Mehta, Kidanu, Verani and Tewolde2015; Dhar et al., Reference Dhar, Jain and Jayachandran2022; Alderman et al., Reference Alderman, Gilligan, Hidrobo, Leight, Mulford and Tambet2025). This is coherent with the idea that norms are shared normative beliefs that require a collective process to be changed. Third, engaging both men and women, either in the same group-discussions (Doyle et al., Reference Doyle, Levtov and Barker2018; Dhar et al., 2022 Reference Dhar, Jain and Jayachandran2022) or by making women participate in related program activities (Alderman et al., Reference Alderman, Gilligan, Hidrobo, Leight, Mulford and Tambet2025), appears to be an important aspect of successful interventions, as this possibly reinforces the message that surrounding social institutions care about gender inequality, updating the perceived social norms. Finally, one-size-fits-all interventions are rarely effective. Curricula must be tailored to local contexts and grounded in a sound theory of change. Evidence highlights the need to challenge dominant ideals of masculinity and gender roles, recognize similarities between men and women, and critically examine men’s privilege and power to align it with women’s rights and empowerment in the economic, political, and domestic spheres (Jewkes et al., Reference Jewkes, Flood and Lang2015).

6.1.4 Correcting Misperceptions

People often tend to assume that others’ behavior is consistent with their attitudes and preferences. By observing a widespread behavior in compliance with the social norm, people will infer that their reference group endorses the norm. When many people privately condemn a behavior but wrongly believe their peers endorse it, we are facing pluralistic ignorance (Bicchieri, Reference Bicchieri2017). In the absence of transparent communication among individuals, people may conceal their true opinions and align with the observed behavior to avoid being put at disadvantage. This may lead to the persistence of a behavior that most people dislike. Child marriage, FGM, and domestic violence can be the object of social norms subject to pluralistic ignorance.

Pluralistic ignorance occurs when people have wrong normative expectations (what they think their peers expect from them), whereas misperception of social norms occurs when people have wrong empirical expectations (how prevalent a certain behavior is). In the case of misperception, for instance, people may (wrongly) think that bribing is common in their country, and adapt their behavior based on this (false) empirical expectation. Whereas in the case of pluralistic ignorance, for instance, families decide not to cut their daughters without telling anyone because they think the norm is to cut the daughters and they would be ostracized otherwise. Both can be solved by either updating people’s normative or empirical expectations, and this can help change behavior. Informing people about the actual frequency of a specific behavior can combat the original misperception of how frequent a behavior really is.

Misperceptions about others, being it in terms of their characteristics – such as race, income, tax evasion – or of their opinions – such as political ideology, gender equality, climate change – are widespread across the world. A review by Bursztyn and Yang (Reference Bursztyn and Yang2022) shows that, out of 81 studies eliciting 434 beliefs, in only 20% of the beliefs the share of respondents who hold correct perceptions about others (i.e., within a 0.5 standard deviation of the truth) exceeds 50%. In more than 30% of the beliefs, more than three-quarters of the respondents hold beliefs that are at least a 0.5 standard deviation away from the truth.Footnote ³⁷ Similarly, misperceptions of others’ beliefs (second-order beliefs) are widely present across the world male and female population when it comes to supporting policies for gender equality in the labor market. Bursztyn et al. (Reference Bursztyn, Cappelen, Tungodden, Voena and Yanagizawa-Drott2023) show that worldwide people underestimate others’ support for basic rights policies, such as allowing women to work outside of the home, by an average of 20.6 percentage points. In 49 out of 60 countries this misperception concerns more than 70% of the population. The pattern is more mixed when it comes to beliefs about other’s support for affirmative action policies, such as prioritizing women when hiring for leadership positions. This appears to be overall underestimated by 9 percentage points, though it is mostly underestimated in less gender-equal countries while it is overestimated in more gender-equal countries.

As mentioned, providing truthful information about others can be a strategy to correct misperceptions. The review by Bursztyn and Yang (Reference Bursztyn and Yang2022) documents that, overall, updating people’s beliefs proves to be effective in modifying their opinions; however, it is much less effective at modifying related behaviors, either actual or intended ones. Indeed, updating empirical expectation will be pertinent only if behaviors are governed by descriptive norms, but not by normative social norms (Bicchieri, Reference Bicchieri2017). Since a social norm is supported by normative expectations, it may be not sufficient to publicly disclose that most individuals dislike the norm and would like to behave differently. To change behavior, people must also be sure that its abandonment will not be followed by negative sanctions.

The literature on updating people’s empirical expectation covers a variety of topics – from peers’ voting participation to water and energy use, from savings to charitable donations (see the review by Haaland, Roth, & Wohlfart, Reference Haaland, Roth and Wohlfart2023) – but very few studies have looked at correcting misperceptions about gender social norms. A study by Bursztyn, González, and Yanagizawa-Drott (Reference Bursztyn, González and Yanagizawa-Drott2020) is a notable exception. Drawing from a sample of 1,500 young men (aged 18–35 years old) living in Saudi Arabia, the authors find that 82% of them are in favor of women working outside of the home, but 92% of them underestimate others’ support. The study deploys an information experiment with a sample of 500 young men in Riyadh, whereby the true rate of support is revealed to one-half of the study participants. These are then asked to make an incentivized choice between receiving an online gift card and signing their wives up for a job-matching mobile application. In the group receiving the updated information, sign-up for the job application increases by 9 percentage points (36% relative to the control group average). Importantly, the authors document long-lasting changes in opinions and actual behaviors.Footnote ³⁸

Overall, there is abundant scientific evidence showing that people widely misperceive others’ beliefs and behaviors, holding concrete consequences for one’s own behavior. Recent evidence also documents that perception of social norms plays an important role in gender equality, especially female employment outside of the home. Correcting those misperceptions is a low-hanging fruit for relatively inexpensive policies and it appears to be effective at changing beliefs and behaviors, at least related to women’s work outside of the home. More evidence is needed to assess the effectiveness of this type of interventions in different contexts and for other gender social norms.