Feminism, Women, and Children

Before 1970, most international human rights organizations focused on state-instigated human rights abuses, such as torture. Abuses that occurred in the home were generally viewed as outside the law and certainly outside the jurisdiction of the international system. For example, when a coalition of African women asked the World Health Organization (WHO) to take action against female genital cutting in 1962, the WHO refused, responding that female genital cutting was a cultural matter (Reference BoyleBoyle 2002). The private sphere was a sacred space that preserved local culture (Reference MerryMerry 2006), and male heads of households were typically perceived as responsible for their household's well-being family law consequently was not a core concern of governments (e.g., McBride & Reference McBride and ParryParry 2011), let alone a concern of international law.Footnote ¹

In fact, global legal regulation of the private sphere has historically been seen as intrusiveFootnote ²; and understanding the shift toward viewing such regulation as necessary requires an understanding of women's global activism. At the turn of the nineteenth century, feminist activists mobilized globally to gain the right to vote (Reference Ramirez, Soysal and ShanahanRamirez et al. 1997). They justified women's formal entry into the political system by arguing that a unique feminine temperament would facilitate peace among countries and that women's roles as mothers and wives would lead to greater attention to children (Reference BerkovitchBerkovitch 1999). For many feminists at the time, women and children's interests were inextricably connected. As Reference MinowMinow (1986: 3) notes: “… the historic roles played by women as advocates for children's needs and interests closely link women's own political struggles with political reforms on behalf of children.”

The feminism that emerged in the post World-War II era in Western Europe and the United States had different foci. Many of these activists de-emphasized women as mothers and generally wanted states to treat women as individual rights holders equal to men (Reference BerkovitchBerkovitch 1999).Footnote ³ After making inroads against gender discrimination in public organizations, many of these activists turned toward violence against women, characterizing it as a universal phenomenon that women disproportionately experienced everywhere, including in the home (Keck & Reference Keck and SikkinkSikkink 1998). Prominent activists and policy makers successfully argued that states and other actors in the global system needed to become more involved in combatting abuses that occurred in the private sphere (e.g., Reference CoomaraswamyCoomaraswamy 2000), aruging that “the private is political.” Ending violence against women was both prominently featured in the 1979 United Nations Convention for the Elimination of All Forms of Discrimination Against Women (CEDAW) and in subsequent mobilizations throughout the 1980s.

This activism transformed a divide between the public and private spheres by asking states to intervene in cases of domestic violence. Once global actors turned their attention to violence against women in the home, it was conceivably only a matter of time before violence against children also became an important global issue.Footnote ⁴ Soon, activists called for the recognition of child rights and the elimination of violence against children, and Sweden became the first country to ban all forms of corporal punishment in 1979 (Reference PaintalPaintal 2007). A decade later, the CRC was open for ratification. The CRC extended many of the rights enshrined in other human rights treaties to children and is the most ratified treaty in the world, as all United Nations memberFootnote ⁵ countries except the United States have ratified it.

Despite its success in prompting international and national legal reforms, much of the Western feminist rhetoric of this period was exclusive and hierarchical (Kapur & Reference Kapur and CossmanCossman 1996), and many Western activists adopted a discourse of saviors and victims that treated women from low- and middle-income states as a homogeneous group (Reference MohantyMohanty 1984). Subsequently, a new generation of academics and activists—led by women from low- and middle-income countries and minority women in wealthy countries—began to focus on intersections of inequality (see, e.g., Reference Budgeon, Gill and ScharffBudgeon 2011; Reference Mack-CantyMack-Canty 2004; Reference MahmoodMahmood 2011). They argued that recognizing rather than eliding differences among women would lead to deeper understandings of inequalities and more effective approaches for combatting them (Reference Buss and BoydBuss 1997; Reference CollinsCollins 2000).

For example, Reference CrenshawCrenshaw (1991) suggested activists should not aim to eliminate all gender or racial differences but rather should recognize these differences as potential sources of social empowerment. In terms of the public/private spheres, this required recognition that, for some women, “the home is not simply a man's castle in the patriarchal sense, but may also function as a safe haven from the indignities of life in a racist society” (Reference CrenshawCrenshaw 1991: 1257). These scholars raised new sensitivities about using the law to address the ills of the private sphere, and the identity politics they fostered may have influenced policies concerning violent discipline.

Taken together, women's global activism informs several hypotheses regarding countries’ adoption of anti-corporal punishment policies. First, an initial emphasis on women's unique concerns as mothers suggests a lasting policy connection between women and children. Scholars argue that men and women legislators make decisions according to gendered interests (Reference KittilsonKittilson 2008; Reference SwersSwers 1998; Reference ThomasThomas 1991) and have found greater representation of women in political bodies affects public spending and the adoption of policies beneficial to women (Reference CelisCelis 2006; Reference Htun and PiscopoHtun & Piscopo 2010; Reference Schwindt-Bayer and MishlerSchwindt-Bayer & Mishler 2005; but see Reference BurnetBurnet 2008; Reference Devlin and ElgieDevlin & Elgie 2008).Footnote ⁶ Despite this, comparatively few studies have considered whether the presence of women in politics is associated with explicitly child-oriented policies or outcomes (but see Reference SwissSwiss et al. 2012). We suggest that women's political representation may positively influence the adoption of policies benefitting children. Put formally:

Nevertheless, there is some reason to believe that Hypothesis 1 will not be supported, as it would be a mistake to assume that all women support violent discipline bans. For instance, a study reviewing attitudes toward violent discipline in nine countries found that some women who believe corporal punishment is unnecessary nonetheless engage in the practice (Reference LansfordLansford et al. 2010), and evidence suggests that many parents, including women, use physical discipline at some point (United Nations General Assembly 2006). It is thus important to test empirically whether there is a connection between women in parliament and child-oriented anti-violent discipline policies.

Second, in line with activists’ emphases outlined above, countries began to protect women's rights in terms of women's public statuses as voters or workers before their private rights, such as the rights to contraception and divorce. Protections from violence followed a similar pattern. For instance, legal protections for rape victims preceded the recognition of domestic violence as a crime. We expect to see a similar pattern for the violent discipline of children, specifically that it will be prohibited first in the public sphere followed by the private sphere. Therefore, we hypothesize:

In line with this, we anticipate the recognition of children's public rights will operate as a precursor to private rights. This is consistent with policies concerning women's rights. Further, policy makers in countries that passed anti-corporal-punishment laws for schools will have likely laid the foundation for laws related to the home. NGOs and activists may also target policy makers in these countries, viewing them as promising reformers because of their earlier actions. We consequently hypothesize:

Finally, we consider whether greater numbers of minorities influence resistance to using law to combat violence in the home.Footnote ⁷ Anti-corporal punishment policies may be less likely in more diverse countries because members of minority groups may promote alternative approaches that rely less on state or police authority. As noted, critical scholars in recent decades have increased awareness of the prejudicial enforcement of such laws and have increased leeriness of policies that invite the state into the home. We therefore hypothesize:

Global Influences on Policymaking

Beyond the role of gender and feminism in the passage of anti-corporal punishment policies, theories of globalization suggest additional factors that affect the worldwide diffusion of laws. Globalization scholars tend to focus more on the process of legal diffusion than on the substance of policies themselves. In general, these scholars fit into two groups: those who emphasize resources and those who emphasize global cultural influences. With respect to the former, a growing body of research considers the relationship between foreign aid and human rights, often reaching conflicting conclusions. Some studies find that aid is associated with human rights and democracy (Reference HearnHearn 1999; Reference PoePoe 1990; Reference TavaresTavares 2003), while other studies find no association (Reference CrawfordCrawford 1997; Reference KnackKnack 2004; Reference MoyoMoyo 2009; Reference Yoo and BoyleYoo and Boyle 2015).Footnote ⁸ In the case of anti-corporal punishment policies, it is difficult to determine the extent to which donors actually link their aid to actions against corporal punishment. On the one hand, donors may consider other rights more important and therefore focus their energies elsewhere, and scholarship has shown that other factors—like political alliances and colonial ties—drive the allocation of aid (Alesina & Reference Alesina and DollarDollar 2000). On the other hand, donors may prefer recipients who enact global norms, and anti-corporal punishment policies may be considered part of the package of global norms (Barrett & Reference Barrett and TsuiTsui 1999). We nonetheless frame our hypothesis in the positive:

Turning to global culture, countries often look to other states or international organizations for ideas about policy reforms (Reference DobbinDobbin et al. 2007). World polity theorists stress that states are part of a global society that shapes the construction of interests and provides clues regarding what “good” states should do (Finnemore & Reference Finnemore and SikkinkSikkink 1998; Reference MeyerMeyer et al. 1997). In fact, scholars have specifically noted the importance of global civil society for promoting child rights (Boyle & Reference Boyle and KimKim 2009; Reference Gran and AlibertiGran & Aliberti 2003; Reference Kim and BoyleKim & Boyle 2011; Reference KimKim et al. 2013).

Global norms may be spread to countries in several ways. First, they may be disseminated through international treaties that delineate global norms countries are expected to incorporate into their national laws. As noted above, the CRC is the core human rights treaty pertaining to children. Although the CRC does not explicitly prohibit corporal punishment, it does prohibit torture and any other violence against children. The CRC Committee also began explicitly discussing corporal punishment in its recommendations to countries in the mid-1990s. For instance, a 1995 Concluding Observation issued by the CRC CommitteeFootnote ⁹ concerning Sri Lanka lamented that there was still corporal punishment in schools and in the home and suggested Sri Lanka take measures to combat the abuse. Several documents created by the Committee solidified this perspective into formal doctrine. A UN study on violence against children was published in 2006 (United Nations General Assembly 2006). This was quickly followed by a General Comment highlighting states’ obligations to prohibit and eliminate corporal punishment of children: “Addressing the widespread acceptance or tolerance of corporal punishmentFootnote ¹⁰ of children and eliminating it, in the family, schools and other settings, is not only an obligation of States… It is…a key strategy for reducing and preventing all forms of violence in societies.” (Committee on the Rights of the Child General Comment 2006). Thus, as the CRC Committee specifically addresses corporal punishment, we anticipate that the CRC will be especially relevant to policy adoption:

INGOs may also influence policy adoption. Indeed, many world polity theorists have demonstrated the link between INGOs and policy adoption, noting their role as carriers of global society (Boli & Reference Boli, Thomas, Boli and ThomasThomas 1999; Reference Frank, Hironaka and SchoferFrank et al. 2000; Reference HironakaHironaka 2002). INGOs often invoke normative arguments to pressure governments to improve their human rights practices through policy reform (Tsutsui & Reference Tsutsui and WotipkaWotipka 2004; Reference Wotipka and TsutsuiWotipka & Tsutsui 2008). The case of corporal punishment is no different, and several international organizations and networks—such as the Global Initiative to End All Corporal Punishment of Children (GI)—have worked closely with the UN to emphasize outlawing corporal punishment. We therefore hypothesize:

Note that GI, which was founded in 2001, has been especially important in accelerating the rate of corporal punishment prohibitions (Reference BowerBower 2015). Because the GI was founded so recently and its founding coincides with several initiatives by UN entities, however, we cannot assess its independent effect in our models; we must rather examine INGOs more broadly.Footnote ¹¹

In sum, we hypothesize foreign aid, the ratification of the CRC, and membership in INGOs will influence the adoption of corporal punishment policies. Recent research suggests that numerous global influences impact national policy adoption and related human rights practices (Kim & Reference Kim and BoyleBoyle 2011) and has linked membership in INGOs and aid (Swiss & Reference Swiss and LonghoferLonghofer 2016). We therefore do not treat these hypotheses as mutually exclusive. It is also important to note that we anticipate that the same factors that influence the adoption of the first policy will influence the adoption of the second policy. In other words, controlling for the adoption of a school policy, these factors will continue to be influential on the odds of adopting a home policy.

Methods and Measures

We conduct the analysis between 1950Footnote ¹² and 2011. This includes as many years as possible, as international data are scarce before 1950.Footnote ¹³ We begin our analysis in 1950 because this coincides with the emergence of global human rights norms and international laws. Specifically, the Universal Declaration of Human Rights was adopted in 1948, and UN commissions completed drafts of the International Covenant on Civil and Political Rights and the International Covenant on Economic, Social, and Cultural Rights in 1954. This time span allows us to examine the factors associated with the lack of policy adoption in early years, which is important for understanding the factors that are associated with policy adoption.

We sought to include all countries as we are interested in general international trends and as all countries could feasibly pass such policies. Ultimately, however, several small nationsFootnote ¹⁴ are excluded because there were not data available for most independent variables. Specifically, there are 150 countries and 5,363 country-years in analyses of school policies, and there are 159 countries and 6,254 country-years in analyses of home policies. The unit of analysis is the country-year,Footnote ¹⁵ and the values vary annually. As with any quantitative study covering many years and numerous countries, this analysis consequently sacrifices locally specific nuance in favor of analyzing general relationships.

Independent Variables

We obtain independent variables, further explained in Table 1 and summarized by policy adoption in Table 2, from a variety of sources.

Table 1. Dependent and Independent Variables

Table 2. Mean Values by Policy Adoption in 159 Countries

^a Aid data, INGO data, and GDP per capita are logged in the analysis.

Women in Parliament

To assess the effect of women in parliament, we include the percentage of legislative seats held by women in the national parliament in each country-year. We focus on national parliaments due to our emphasis on countrywide bans, as opposed to regional bans, on corporal punishment. These data were obtained from Reference PaxtonPaxton et al. (2006) as well as from the International Parliamentary Union (2014).Footnote ²⁰ As we explain in more detail below, we also test a threshold effect of 30 percent or more seats belonging to women.

Ethnic Fractionalization

Plurality within the population is operationalized with an indicator of ethnolinguistic fractionalization, which measures the probability that two randomly drawn individuals from a country are from different linguistic groups (Fearon & Reference Fearon and LaitinLaitin 2003). Constructing any comparative measure of ethnic groups across countries raises validity concerns. Fearon's ethnolinguistic fractionalization measure gives special attention to linguistic diversity, which is not perfectly correlated with ethnic diversity but is better than alternate measures.Footnote ²¹ Fearon (2003: 210) does take into account factors other than a shared language, however, such as indicators of a common culture or race. He constructed this metric using data from the Atlas Narodov Mira 1964, which provides the best-known and most widely used measure of ethnic fractionalization. He then supplemented the Atlas with data from the CIA World Factbook, the Encyclopedia Britannica, the Library of Congress Country Studies, and other sources in order to take other indicators of culture into account. As this measure remains quite constant over time, we interpolated it for missing years.Footnote ²²

International Aid

Given our focus on child rights, we include a measure of civil society aid from AidData (2016; see also Reference TierneyTierney et al. 2011). Civil society aid includes aid given for human rights, women's groups, grassroots organizations, and other efforts that promote civil society. AidData provides the raw amount of aid given in USD 2011 across all country-years; we thus added all aid allocated to each country for each country-year and logged this value to best fit the data.

CRC Ratification

As the CRC Committee monitors corporal punishment and as human rights treaties reflect international norms, we include a measure of ratification of the CRC.Footnote ²³ We operationalize ratification as the number of years since ratifying the CRC, which was open for ratification in 1989.

INGOs

INGOs are operationalized as the number of INGOs to which any citizen of a country belongs. The data are obtained from the Yearbook of International Organizations (various years) and are logged. As is consistent practice, data for earlier periods are interpolated (see Kim & Reference Kim and BoyleBoyle 2011).

Analysis

To assess the factors that influence whether and when countries adopt corporal punishment policies in the home and in schools, we utilize discrete-time hazard models and, again, model the adoption of policies in the home and in the school separately. The hazard is the instantaneous propensity that an event will occur. Unlike many models, hazard models analyze the influence of time and allow for time-varying predictors and censoring of data. In other words, countries are no longer considered “at risk” of adopting a certain policy once policy adoption has occurred, as there are no cases of legislation being overturned.Footnote ²⁷

Specifically, we employ discrete-time logistic regression models (Allison 1984):

$log\left[P_{it}/\left(1- P_{it}\right)\right] = {\alpha }_t+ {\beta }_1{X}_{it1}+ \dots + {\beta }_k{X}_{itk}.$

In this model, P_it represents the probability that country i enacts a policy at time t. β signifies the effect of the independent variables; X ₁, X ₂ … X_k denote k time-varying explanatory variables; and α represents a set of constants corresponding to each discrete-time unit of one country-year. We also cluster by country identifier to adjust for correlated errors within countries over time.Footnote ²⁸

The discrete-time hazard model invokes several assumptions. First, it invokes linear additivity, which means that a predictor's effect does not depend on the values of other predictors in the model or the position of the unit difference along its scale. Second, it invokes the proportionality assumption: that each predictor has an identical effect in every time period under study. In other words, it does not depend on the duration. Finally, the discrete-time hazard model invokes an assumption that there is no unobserved heterogeneity. Essentially, the population hazard depends only on predictor values (Singer & Reference Singer and WillettWillett 2003).

Analyses were conducted in Reference StataCorpStata (2014), and results for the two models are presented in odds ratios with exponentiated confidence intervals. Coefficients larger than one are associated with increased odds of a policy, while coefficients smaller than one are associated with decreased odds of having a policy. Again, we cluster standard errors by country identifier, though the Supporting Information Appendix includes random effects models as a robustness test. Results remain qualitatively similar, providing additional support for the results of our event history analysis.

Results: Banning Corporal Punishment

Figure 1 shows the number of countries that adopted corporal punishment policies in the home and in schools between 1950 and 2011.Footnote ²⁹ As the figure clearly illustrates, few countries banned either form of corporal punishment during the first few decades of our analysis. Instead, a handful of countries sporadically adopted policies until 1990, when school policy adoption increased dramatically. Home policy adoption also greatly increased during the 1990s, though most policy adoption occurred later in the decade. Overall, the 1990s and 2000s saw the largest rise in the adoption of each form of policy, with the highest number of policy adoptions occurring between 2005 and 2011.

Figure 1. Number of Corporal Punishment Policies Adopted Globally, 1950–2011. Note: Only new policy adoptions are shown; thus, the figure does not show cumulative policy adoption.

This figure illustrates important general trends and supports our hypothesis that public (school) policies will be adopted prior to private (home) policies (Hypothesis 2). In 1950, sevenFootnote ³⁰ countries had previously adopted a policy banning corporal punishment in schools, and no countries had policies outlawing corporal punishment in the home. By 2011, over 70 countries had banned corporal punishment in schools, and over 30 countries had outlawed corporal punishment in the home. This trend—increasing numbers of policies being adopted over time—suggests a norm cascade (Finnemore & Reference Finnemore and SikkinkSikkink 1998).

While we focus on the two forms of policies as indicators of a similar phenomenon, these trends illuminate differences between the two types of policies. On a global scale, countries adopted policies pertaining to schools well before they adopted policies pertaining to the home. Policies regarding schools are also more than two times as prevalent as policies regarding the home. In fact, only one country (Turkmenistan) has adopted a policy banning corporal punishment in the home but has not banned corporal punishment in schools, and only one other country (Iceland) adopted a policy pertaining to the home before adopting a policy pertaining to schools. These trends suggest that child-friendly policies related to the public sphere—in this case, educational institutions—are more prevalent than policies pertaining to the private sphere of the home. To understand these processes in more depth, we now turn to a multivariate analysis.

Multivariate Analysis

Table 3 presents discrete time hazard models of the adoption of the two forms of policy. Models 1, 2, and 3 pertain to school policies, while Models 4, 5, and 6 pertain to home policies. Model 7 considers home policies when the earlier adoption of a school policy is controlled, and each of the seven models also includes controls for the type of government, development, and time, as explained above.

Table 3. Discrete Time Hazard Models of Adoption of Corporal Punishment Policies, 1950–2011, Odds Ratios

Standard errors are clustered by country identifier, and exponentiated confidence intervals are included in parentheses.

^***p < 0.001, ^**p < 0.01, ^*p < 0.05, ⁺0p < 0.1.

Abbreviations: Ethfrac = ethnolinguistic fractionalization.

Hypothesis 1 suggested that the percentage of women in a country's national parliament is positively associated with the adoption of each form of policy. As seen in Model 1, the percentage of women in parliament is associated with the adoption of policies against violent discipline in schools. Each increase in the percentage of women in parliament is associated with a 4 percent marginal increase in the odds of adopting a school policy (p < 0.01). The percentage of women in parliament is also significantly associated with the odds of banning corporal punishment in homes (p < 0.001) as seen in Model 4, and this effect is particularly robust across models.Footnote ³¹

The significance of the percentage of women in parliament falls in line with existing research that finds that having women in parliament is associated with a host of outcomes related to women's and children's well being. The relative strength of this effect on banning corporal punishment in the home also suggests that women may be more effective at mobilizing change related to the home—a realm traditionally associated with women—than in educational institutions, which may be considered the domain of men in certain countries.

We also hypothesized that more diverse countries may avoid law as a solution to the issue of violent discipline. This hypothesis is based on the idea that members of minority groups may be especially vocal opponents of policies that invite the state into their homes (Hypothesis 4). This hypothesis is supported, as greater ethnolinguistic fractionalization is significantly associated (p < 0.05) with a lower likelihood of adopting an anti-corporal punishment policy in homes (Model 5). The significant interaction effect in Model 6 shows that ethnolinguistic fractionalization is especially influential when there are more women in parliament. Indeed, there are early adopters of home policies—such as Finland, Norway, and Denmark—where ethnolinguistic fractionalization was low and the representation of women in parliament was continually increasing. At the same time, there are ample examples of countries that have high representation of women in parliament, high ethnolinguistic fractionalization, and have yet to adopt policies banning corporal punishment in the home (such as Bolivia, Tanzania, Senegal, and Canada). However, given the large conditional effects when this interaction is introduced (which likely is the result of a relatively small number of home policies), we urge caution in interpreting it and urge future researchers to consider whether the finding is replicable.

Turning to global factors, we also find that aid allocated to civil society pursuits—which includes aid provided for human rights, women's groups, grassroots organizations, and other efforts that promote civil society—is associated with significantly greater odds (p < 0.01) of adopting both policies pertaining to schools (Model 2) and policies pertaining to the home (Model 5). This lends strong support to Hypothesis 5. While this aid may have a direct effect on mobilization efforts, we contend it more likely represents the spread of a global norm against corporal punishment. For example, Theresa Kilbane, Senior Advisor of Child Protection at UNICEF (Reference KilbaneKilbane 2016), revealed in an interview that even within UNICEF's very decentralized structure, in which each country's office has substantial latitude over the focus of its efforts, there has been an uptick in the number of countries choosing to focus on violent discipline. Numerous scholars have found that aid is associated with human rights outcomes, and aid is often allocated under certain conditions (Abrams & Reference Abrams and LewisLewis 1993). Organizations that provide aid also monitor their aid recipients, suggesting close ties to other countries and norms they propagate.Footnote ³²

Turning to other global factors, early ratification of the CRC is associated with higher odds of home policy adoption. Every year since a country has ratified the CRC increases the odds of adopting a policy by a factor of 1.1 (p < 0.001). This partially confirms Hypothesis 6 and illustrates the importance of the CRC. Yet, while CRC ratification is significantly associated with the odds of adopting a policy banning corporal punishment in the home, it is not significantly associated with the odds of adopting a policy banning corporal punishment in educational institutions.Footnote ³³ This difference may be explained by most countries’ hesitancy to regulate behavior in the private sphere. Formal insisting from the CRC Committee thus may have influenced the regulation of behavior in the home, though the relatively small number of countries that have adopted such policies suggests that such insisting is also not sufficient for policy adoption.

Contrary to Hypothesis 7, membership in INGOsFootnote ³⁴ is not significantly associated with the adoption of either form of policy. While this measure is correlated with time, it is also insignificant when time is excluded from the models. Likewise, excluding aid from the models did not bolster the INGO effect. To further probe this finding, we examined the effects of women's INGOs, children's INGOs, and health NGOs, and none was significantly related to either policy adoption. This is surprising because INGOs are consistently associated with the diffusion of policies in other studies of global processes (Reference Frank, Hironaka and SchoferFrank et al. 2000, Reference Frank, Camp and Boutcher2010; Reference SwissSwiss et al. 2012). One plausible explanation is that the ideas of global civil society are being carried through many different networks, attenuating the unique importance of INGO networks (Reference Hironakacompare Hironaka 2014). Another possibility is that pressure for reform from INGOs has become less powerful over time, and we encourage future researchers to explore these possibilities further.

As noted above, descriptive statistics indicated that almost all countries that had adopted a home policy had previously adopted a school policy. To test this in our multivariate analysis, Model 7 includes a dichotomous variable indicating whether a country had adopted a school policy at least one year prior to the country-year.Footnote ³⁵ As seen in the table, previously adopting a school policy is the strongest predictor of enacting a home policy, which provides support for Hypothesis 3. While weaker, the effects of the other key independent variables on home policies remain when school policies are controlled—women in parliament, aid, and ratifying the CRC are significant, while INGO connections are not. This indicates that similar factors are associated with policy adoption regardless of whether it is a first or a second policy concerning violent discipline.

Finally, control variables yield several findings. GDP per capita is significantly associated with home policy adoption, indicating that wealthy countries are more likely to extend human rights regulations into the home. Yet, GDP per capita is not associated with the adoption of policies that ban corporal punishment in educational institutions, which have been adopted in a wider range of countries. This may indicate that developed and developing countries have somewhat different views on whether corporal punishment in the home is harmful. Conversely, democracies have higher odds of adopting policies pertaining to schools but not to the home. Democracies may hesitate to regulate certain elements of the domestic sphere because such regulation may be viewed as infringing on the rights of their citizens. This also suggests that autocratic regimes may treat education as an institution that encourages compliance and obedience.

Conclusion

Previous research has demonstrated negative consequences of corporal punishment, such as lower math scores and less engagement with school (Reference OgandoOgando et al. 2015). Our research reveals a positive development—that formal condemnation of corporal punishment is becoming a global norm. While few countries had banned corporal punishment in either schools or homes during the first few decades of our analysis, Figure 1 illustrates that the 1990s saw a large increase in policy adoption. Policies banning corporal punishment in schools have proliferated widely, while policies banning corporal punishment in the home are being adopted more slowly. School policies are a strong predictor of home policies, and adopting a home policy prior to a school policy is very rare. These trends mirror the public/private pattern of women's rights reforms, suggesting that state institutions are relatively more permeable than the institution of the family.

In line with this, our event history analysis of more than 150 countries between 1950 and 2011 suggests a nexus between women's and children's issues. Women in parliament are associated with the adoption of anti-corporal punishment policies in both schools and homes. Our findings are also consistent with the idea that protecting children from violence is part of the general anti-violence framework that feminist activists have promoted. Most other forms of private violence have been addressed in legislation, and activists and policy makers have now turned their attention to creating norms against violent discipline (Reference KilbaneKilbane 2016). Nevertheless, this finding is not completely obvious. At least one study, based on surveys in nine countries, found that women on average were no more opposed to corporal punishment than men (Reference LansfordLansford et al. 2010).

We also find that home legislation is less likely in countries with high levels of ethnic diversity. In countries with diverse populations, members of minorities may support alternative ways of dealing with violent discipline because they are concerned about the consequences of greater policing of the private sphere. To be clear, this does not indicate that members of these communities support violent discipline or that support for violent discipline varies by ethnicity. Rather, it highlights the potential reluctance of minority communities to support policies that bring the state into their homes. We find that this effect is especially pronounced in countries that have more women in parliament, which could mean that elite women are particularly sensitive to the concerns raised by critical theorists. Unfortunately, there are no public data on the ethnic make-up of legislatures, so it is not possible to test this directly.

Our findings also have implications for understanding the global diffusion of policies. Most notably, the receipt of international aid is tied to the adoption of both school and home policies. It seems unlikely that there is a direct quid pro quo relationship between corporal punishment policies and the receipt of foreign aid; there are likely many other human rights-related policies that garner more attention. Rather, this finding suggests that countries that signal respect for global reforms and international law generally are favored by donor states (see Barrett & Reference Barrett and TsuiTsui 1999). Anti-corporal punishment policies are relatively easy to adopt, in part because their enforcement is difficult to measure.

With respect to indicators of global norms, CRC ratification does not predict school policies. Many of these policies precede or closely coincide with CRC ratification. Our finding may indicate that a global norm opposing corporal punishment prompted both the ratification of the CRC and the adoption of the policy within countries, as there was already a high degree of consensus around this issue by the time the CRC was opened for ratification. This is consistent with Reference HironakaHironaka's (2015) “swarm” metaphor for the spread of global norms. At the same time, there is a strong connection between the ratification of the CRC and the adoption of policies banning corporal punishment in the home. This implies that the CRC put states under pressure to expand child rights into the more controversial private sphere.

More INGO memberships signal dense ties to a global polity that shapes norms and fuels reforms. Yet, links to INGOs surprisingly have no significant association with the adoption of anti-corporal punishment policies in our models. This may be a consequence of the profound and rapid embrace of child rights in the immediate post-Cold War era. The CRC was ratified more quickly and by more countries than any other global treaty. Further, as noted above, many school policies preceded the CRC. Together, this implies more consensus—at least at the policy level—than is typical for global reforms. Our finding suggests that even states that are less connected to the world polity did not need much prodding to adopt policies against violent discipline. The lack of an association of policies with INGOs, in combination with the strong association of policies with women in parliament, could also mean that states rather than civil society have been the primary movers of corporal punishment policy adoption.Footnote ³⁶

At any rate, INGOs may play a vital role in holding countries accountable for the policies they have adopted (Hafner-Burton & Reference Hafner-Burton and TsutsuiTsutsui 2005). The enforcement of policies differs from their adoption. In the specific case of corporal punishment, evidence suggests that many of the countries that adopt laws do not enforce them (World Health Organization 2014). At the same time, the recently-launched Global Partnership to End Violence Against Children (2016), which brings together state and non-state organizations in the fight to end violence against children, cites policy enactment as one of its cores strategies. We hope that future studies will take up the question of how the actual practice of violent discipline is changing over time. Indeed, ongoing data collection by UNICEF (2015) through the multi-indicator cluster surveys is paving the way for future analyses of how policies are affecting behavior in this sphere.

As this was the first study to assess the factors associated with policies banning corporal punishment in the home and in schools, this quantitative approach has provided a broad foundation of knowledge regarding the general factors that are associated with policy adoption. Going forward, scholars should seek to refine the relationships tested here and to assess the precise mechanisms behind them. Further, change in spheres other than law, such as education or media, must also be considered. Violence against children in schools and in homes remains a pressing social problem, and sound sociological and legal research is vital in better understanding corporal punishment and global efforts to address it.