Worldwide violence against women is a major public health and human rights problem, according to the World Health Organization, (2002). This sad situation is compounded by the ethical, safety and methodological concerns involved in researching this problem (WHO, 2003). Underreporting cases illustrates a methodological concern. Despite limitations, existing research has explained violence against women based on class, age, race, religion, cultural, social relations, educational and occupational status, and institutional explanations (Ellsberg et al., 2001; Yodanis, 2004; Sokoloff & Dupont, 2005; Flood & Pease, 2009). Scholars also have tested the explanatory power of gender representation (Miller & Segal, 2019) on reported cases of violence against women.

In 2008, the United National Secretary-General launched a multi-year global campaign called UNiTE to End Violence against Women. The campaign recognizes the power of the law. One of its five key goals was for all countries to adopt and enforce national laws and institutions that address and punish all forms of such violence, in line with international human rights standards (United Nations, 1993). The promoted laws were targeted to criminalize such violence, ensure the prosecution and punishment of perpetrators, empower and support victims, and strengthen prevention. In response to this global campaign, countries have opted for promoting the creation of pro-women institutions, with the goal of empowering and supporting victims and strengthening violence prevention. Brazilian municipalities, for example, were encouraged to create several pro-women institutions, such as a municipal secretary for women’s rights, women’s police stations, reference centers for women, and shelters for women facing violence (Brazil, 2011; Calazans & Cortés, 2011; Pardo & Sanematsu, 2017; Nunes, 2017; see also Fig. 1). However, we know little about the effects these pro-women institutions have had on reporting cases of violence against women.

Fig. 1
figure 1

Existing Brazilian laws to protect women against violence. Source: prepared by the authors based on data from Senado Federal (2019)

Consequently, this study pays particular attention to the role of pro-women institutions on the effects of reporting violence against women. Two reasons motivated our focus on pro-women institutions. First, unlike economic, demographic, and contextual factors, pro-women institutions are instruments available for adoption by any country and any subnational government. Second, despite the worldwide call for adopting pro-women institutions, considerable variation exists in such institutions both across countries and within countries, giving researchers a justifiable reason to test the police reporting effect of this variation. Finally, exploring the benefits of pro-women institutions also should encourage researchers to investigate the effects of other pro-minority group institutions (e.g., pro-refugees, pro-gender, pro-indigenous, pro-ethnic, pro-minority religious groups, etc.).

Specifically, the study derives data from the 496 municipalities of the Brazilian state of Rio Grande do Sul during two different periods, 2013/2014 and 2017/2018. The considerable cross-municipal variation in the number of pro-women institutions, data availability, and the slightly higher homicide rate against women (5.2 for each group of 100,000 women in 2017 compared to the 4.5 national rate, IPEA, 2019) make this state worthy of study. Data on violence against women are from 2014 and 2018, and data on testable explanations are from 2013 and 2017. Brazil is a semi-industrialized country, whose municipal contexts and legislation in favor of pro-women institutions also make it worth studying. Brazil exhibits one of the highest rates of violence and murder against women in the world, according to data from the Women Stats Project ( The violence persists despite the country adopting many laws and institutions seeking to protect women from domestic and general violence. This paradox calls for empirical evidence on violence reports of pro-women laws and institutions.

Violence against women is captured with reporting of five different violent actions: (a) serious threats, (b) personal injury, (c) rape, (d) attempted murder, and (e) murder. We focus on the reporting effects of having seven specific pro-women institutions: (i) a municipal secretary for women’s rights, (ii) a city council for women’s rights, (iii) a women’s police station, (iv) a special court of violence against women, (v) a women’s public defender office, (vi) a reference center for women, and (vii) a shelter for women facing violence. The study controls for municipal-level characteristics, including demographic, economic, social, and cultural features, as well as for female representation in political (mayoral and city council) and bureaucratic (police chief) posts.

Results seem to be contingent on the type of reporting case and pro-women institution. Having a police station for women’s issues boosts reporting cases of personal injury, and the presence of a public defender officer for women increases reporting cases of serious threats and attempted murder.

Violence against women

The term ‘violence against women’ has been addressed differently in the literature. Some studies emphasize the context of violence (Aizer, 2010; Anderberg et al., 2013), while others stress the specific acts that characterize violence (Watts & Zimmerman, 2002), and still others focus on victims and the consequences of violence (Smith, 1994). According to the United Nations’ Declaration on the Elimination of Violence against Women (UN 1993, p. 2), violence against women means “… any act of gender-based violence that results in, or is likely to result in physical, sexual or psychological harm or suffering to women.” According to the Declaration, ‘gender-based’ violence involves unequal relationships between men and women and emphasizes that violent acts are rooted in sex inequality.

Violence against women includes specific forms of abuse, such as physical violence ranging from slaps, punches and kicks to female genital mutilation, acid throwing, rape, dowry deaths, assaults with a weapon, and murder. It also includes several kinds of sexual violence, such as forced sex or forced participation in degrading sexual acts, and psychological violence (Watts & Zimmerman, 2002). Violence against women by a husband or an intimate partner is usually referred to as ‘intimate-partner violence.’ Garcia-Moreno et al., (2005) defines intimate partner violence as the “physical, sexual and emotional abuse by a current or former intimate male partner, whether cohabiting or not.

In addition to the distinction between violence against women (broader definition) and domestic violence (narrower definition), Kilpatrick, (2004) offers another distinction, which derives from contrasting a criminal justice approach with a public health approach. From the criminal justice perspective, violence against women would be defined as the subset of violent crimes perpetrated against women. This definition excludes psychological abuse (Kilpatrick, 2004). In most countries, the criminal justice approach identifies murder, assault, rape, and stalking crimes, irrespective of the relationship between perpetrator and victim. On the other hand, the public health perspective defines violence against women as a subset of interpersonal violence and includes the typologies of physical, sexual and psychological violence (Kilpatrick, 2004).

In this paper, violence against women is broadly defined following the criminal justice approach. That is, violence against women includes physical and sexual violence crimes perpetrated by a husband, a partner, and/or anyone else. Consequently, our dependent variable captures state police reports of serious threats, personal injury, rape, attempted murder, and murder against women, including children. Our definition excludes psychological abuse because in many cases, it is not categorized as a crime (Kilpatrick, 2004). Therefore, here, violence against women is limited to cases reported to the police. It is important to highlight that although murder reports equal the actual cases that occurred, reported numbers of other violent acts reflect victims’ willingness to report, as non-reported cases are left out.

Previous studies offer some drivers and obstacles to police reporting. For example, severity, property damage, and use of a weapon encourage police reporting (Akers and Kaukinen, 2009; Chen & Ullman, 2010). On the other hand, lack of injury discourages reporting, in part because of fears of disbelief (McGregor et al., 2000). Gender-based and race-based inequities intersected at the structural and community levels discourage women from police contact following intimate partner violence and sexual violence (Chen & Ullman, 2010).

Several innovations have been created to help women overcome obstacles that hinder or even prevent women from reporting aggressors (Jewkes & Dartnall, 2019). An example is digital technologies that guide and provide security mechanisms so that reports can be made more easily. Digital technologies are tools or resources that can be integrated with other effective programs that women can use anonymously, 24/7, without feeling judged. These digital technologies include web pages or mobile device applications, with extensive safety features delivered through appropriate technology (Koziol-McLain et al., 2018; Hegarty et al., 2019; Wood, Glass and Decker, 2019). The next section will address the most common drivers of violence against women.

Drivers of violence against Women

The literature provides several explanations for the occurrence of violence against women (Ellsberg et al., 2001; Yodanis, 2004; Sokoloff & Dupont, 2005; Flood & Pease, 2009; Miller & Segal, 2019). The theoretical models describe both the risk and protective factors of this type of violence based on social, legal, political, cultural, economic, biological, psychological, and gender equality concepts. The most widely used model is called the ‘ecological model,’ which proposes that violence is the result of factors operating at four different levels: individual, relationship, societal and community (Heise, 1998). The United Nations and the World Health Organization have embraced this ‘ecological model’ because it addresses both the factors instigating women to suffer abuse and violence, and the drivers prompting individuals to commit acts of abuse and violence (Garcia-Moreno et al., 2005).

Another important theoretical model used to explain violence against women is the ‘feminist theory,’ which sees violence against women as the result of gender inequality on the societal level (Bograd, 1988). The more unequal women are in a specific society, compared to men, the more likely men are to be violent toward them. The roots of gender inequality are both ideological and structural because beliefs, norms, and values about the status and roles of women in society and women’s access to positions within social institutions determine gender gap (Dobash & Dobash, 1979). According to Gelles, (1993), the feminist theory had become “the dominant model for explaining violence toward women” (41).

In addition, Bradley and Khor, (1993) emphasize the economic, social, and political dimensions of gender inequality. The economic dimension focuses on women’s status in “activities and institutions built around the production, distribution, and consumption of goods and services” (Bradley & Khor, 1993, 349). Economic models focused on individual aspects of domestic violence indicate that women’s greater bargaining power in marriage results from their greater participation and positioning in the labor market, which in turn decreases violence (Cerqueira et al., 2019). Evidence of this assumption is found in studies conducted in several countries (Farmer & Tiefenthaler, 1997; Koening et al., 2003; Markowitz, 2001; Nagel et al., 2005; Bhattacharya, Bedi & Chhachhi, 2009; Grabe, 2010; Aizer 2010; Anderberg et al., 2013).

In addition to the economic dimension of gender inequality, social and cultural contexts in which women live should promote or protect them from violence. Within this dimension, scholars have given considerable attention to race and ethnicity, religion, traditions and cultural patterns, among others (Flood & Pease, 2009; Alhabib, Nur & Jones, 2010). According to Douki et al., (2003), in general, religion supports the ideal model of a patriarchal family where women are under the authority of their husbands, which may encourage violent male behaviors. Although little evidence exists of religion’s impact on violence against women, evidence exits of contexts in which religion is (mis)used to justify violence against women or to perpetuate women’s vulnerability to victimization (Flood & Pease, 2009). This kind of evidence can be observed in almost all religions.

The political context in which women grow and operate is also expected to influence their risks of violence. Political status refers to women’s access to power and representation in the state and societal institutions (Yodanis, 2004). In male-dominated institutions, violence is a tool that men can use to keep women out or subordinate in order to maintain male power and control. When men dominate family, political, economic, and other social institutions both in number and in power, the policies and practices of these institutions are likely to embody, reproduce, and legitimate male domination over women, and that practice will be considered right and natural in these institutions and throughout the society in general (Brownmiller, 1975; Riger & Gordon, 1981; Stanko, 1990; Yodanis, 2004). Finally, to address gender segregations, scholars, practitioners and policy makers have turned to the postulates of the representative bureaucracy theory. According to this theory, passive representation of minority groups leads to their active representation (Krislov, 1981; Mosher, 1982; Thielemann & Stewart 1996; Howell & McLean 2001; Kelly & Newman 2001; Meier & Nicholson-Crotty, 2006; Wilkins, 2006; Wilkins & Keiser, 2004).

Institutional drivers of violence against women

Scholars also have highlighted the role of institutional drivers in explaining violence against women. For instance, adoption and implementation of punitive laws requiring legal actions for the perpetrators of violence against women should have a deterrent effect (Devries et al., 2013; Ellsberg & Emmelin 2014). Deterrence legislation varies in scope and degree. Table 1, for example, shows the percentage of countries in each region of the world that have laws to combat different types of violence against women. The data from a World Bank study conducted in 2016 show there are criminal sanctions and specific laws addressing emotional, economic, physical and sexual violence. South Asian region includes more countries with legislation (85.4%), followed by Latin American countries (69.3%). Middle East and North African countries rank last, as only an average of 12.3% of the countries have adopted laws punishing violence against women.

Table 1 Percentages of countries, by region, with legislation for different types of violence against women

Not all agree on the deterrence effects of punitive laws. Some, for example, contend that having punitive rules is a necessary but not a sufficient condition to deter perpetrators from attacking women (Htun & Weldon, 2012). In Brazil, for instance, violence against women has steadily increased in the last years despite the adoption of many punitive laws (see Fig. 1) (IPEA, 2019). Therefore, we suggest that, to be effective, punitive laws need to be accompanied by local pro-women policies and institutions, thus creating an institutional framework that is able to channel demands and respond to them through specific actions.

Governments can more directly convey a strong commitment toward protecting women by favoring the creation and sustainability of pro-women institutions. The emergence of pro-women institutions has increased and become prominent even in regions and countries with long-standing male traditions (Htun & Weldon, 2012). Pro-women institutions should empower and encourage women to report violent attacks against them (Avdeyeya, 2009; Franceschet, 2010; Htun & Weldon 2012). As pro-indigenous reforms/institutions have empowered and protected indigenous rights (Oliart, 2008), pro-women institutions also are expected to do their part for women. The present work investigates the police reporting effects of different pro-women institutions: (a) political institutions, such as secretariats for women’s rights and municipal councils for women’s rights; (b) justice institutions, such as police stations specializing in violence against women, a special courts and a public defenders officer for women; and (c) support institutions, such as reference centers and shelters for women facing violence. In the following paragraphs, we offer a set of hypotheses for the effects of each of these kinds of local institutions on reports of violence against women.

The creation and permanence of institutions that seek to defend the rights of a specific group should convey members of that group government’s pledge to do so. Local government scholars contend that institutions that spur neighborhood-based political participation help provide voice to minority and underrepresented groups, enhance citizen efficacy, and are integral to a thriving democracy (Stone & Stoker, 2015; Einsten, Palmer & Glick, 2018). For instance, the passage of a municipal women policy and/or establishment of a women’s rights council whose mission involves promoting women’s rights should communicate formal government recognition of women’s rights and states’ commitment to guarantee their individual and collective rights. This municipal commitment, in turn, should boost women’s trust in their local governments and help them perceive governments as their advocate, thus encouraging women to report violence ordeals. Pro-women’s rights institutions also should serve as participatory channels through which women can voice their demands, unequal access, and aggression exposures. Therefore,

H1: Municipalities with institutions promoting women’s rights are more likely to register more cases of violence against women

Likewise, municipal existence of justice/legal/penal institutions aimed exclusively toward women also should encourage them to report violent ordeals. Specifically, presence of a police station, a special court, and a public defender office all exclusively dedicated to deal with violence against women conveys municipal financial and administrative adherence to creating specific venues where their complaints can be heard. Moreover, most of these specialized institutions are staffed with women. Women representation in these institutions may inspire a sense of understanding and empathy to further encourage women to denounce aggression acts. Likewise, having a public defender for women, allows women to have a lawyer regardless their economic status. In addition, justice systems tend to be overwhelmed by the increasing number of cases, leading to delays and sometimes nullification of penalty due to term limits. Justice institutions for women should help decongest penal systems, encouraging women to file complaints with the hope their cases will be addressed soon. Therefore,

H2: Municipalities with justice institutions for women are more likely to report more cases of violence against women

In addition to women pro-rights and justice institutions, municipal existence of supporting institutions, such as shelters and reference centers, also should generate violence reporting effects. The creation of pro-women institutions alone may not guarantee the expected results (Einsten, Palmer & Glick, 2018). In fact, studies suggest that access to support services, medical care, and forensic exams increases police reporting (Zweig et al., 2014; Marchetti, 2012). For instance, reference centers offer legal and psychological support for women in situations of violence, and shelters offer temporary housing and food for women who have suffered domestic violence. Moreover, studies also suggest that experiencing violence from a known aggressor, including partners, discourages police reporting (Chen & Ullman, 2010; McGregor et al., 2000), likely reflecting social or economic dependence, and fear of retribution (McGregor et al., 2000; Marchetti, 2012). Perpetrators sometimes may physically prevent calls to police, and/or manipulate police officers (Wolf et al., 2003). Consequently, municipal existence of these supporting institutions should signal to women they are not alone and can rely on governmental institutions to overcome potential repercussions of reporting suffered aggressions. Not having a safe place to go ties women to tolerate and stay with the perpetrator, discouraging them to fail a complaint. Consequently,

H3: Municipalities with supporting institutions for women facing domestic violence are more likely to report more cases of violence against women

Besides the expected reporting-enhancing effects of each type of pro-women institution, the total number of local pro-women institutions also should have an additive effect. In fact, in certain settings, local governments have the autonomy to create one or several of pro-women institutions. It is a relevant research problem, but the drivers leading to adopting these pro-women institutions are outside of this research’s scope. Although not all pro-women institutions may directly prevent women from being victims of violence, as in the case of police stations for women, the institutions’ goals are to support and encourage women to report violence and abuse (Avdeyeya, 2009). These pro-women institutions rely on media campaigns and personal services to reach out women. Therefore, a higher number of pro-women institutions should empower women to speak out and report their experiences with both domestic and general violence. Women’s trust, perception, and confidence in their local governments should be influenced by the number and type of pro-women institutions, which, in turn, may enter into their risk assessments when considering reporting or not aggression ordeals. Therefore,

H4: The greater the number of pro-women institutions in a municipality, the greater the number of reported cases of violence against women

Research design and data operationalization

The derived hypotheses can be tested in any context and across units of any level of government. In this study, the unit of analysis is the municipality-year. Specifically, the analysis covers all 496 municipalities of the Brazilian state of Rio Grande do Sul in two different periods, 2013/2014 and 2017/2018. Data on violence against women are from 2014 and 2018, and data on pro-women institutions are from a year earlier, 2013 and 2017, respectively, creating a balanced panel data with 992 observations.

We selected the state of Rio Grande do Sul (RS) for several reasons. First, it is one of the most important Brazilian states. RS is the fifth most populous state, with approximately 11 million people (IBGE, 2019). In economic terms, RS has the fourth-largest gross domestic product (GPD), surpassed only by the states of São Paulo, Rio de Janeiro, and Minas Gerais. RS has the largest number of elderly people in the country, the second-highest life expectancy, and one of the highest schooling rates and lowest child mortality rates in the country (IBGE, 2019). As mentioned before, RS’s homicide rates against women closely resembles the national rates. Mortality official data show that RS had a rate of 5.2 homicides of women in 2017 for each group of 100,000 women, slightly above the national average of 4.7. Out of the 27 Brazilian states, the state of São Paulo had the lowest rate (2.2), and the state of Roraima the highest (10.6) (IPEA, 2019).

Brazil has showed a significant growth of 30.7% in the number of homicides of women during the 2007–2017 period, according to data aggregated by IPEA, (2019). In 2017 alone, 4936 women were killed, approximately 13 murders per day, the highest number since 2007. These figures show the magnitude of the problem of violence faced by women in the country, with dramatic social and economic results. Despite the large number of cases of violence against women, Brazil has many laws that seek to protect women from domestic and general violence. The federal government has exclusive control to legislate on this subject, and its laws are valid across all states and municipalities.

The first most significant pro-women legislation in Brazil was enacted in 2003, named Law 10,778. This law requires notifying all cases of violence against women and registering them in the country’s health system. This helps to understand the magnitude the problem and define appropriate public policies to address it. However, the legal framework for combating violence against women in Brazil only emerged in 2006 with the approval of Law 11,340, which sought to curb domestic and family violence against women by punishing attackers. This law is known the ‘Maria da Penha Law’ in honor of Maria da Penha Maia, who was battered for six years by her husband until she became a paraplegic in a firearm attack in 1983. The United Nations considered this law as the third best law against domestic violence in the world in 2012. Figure 1 provides a chronological list of the main Brazilian laws on violence against women. The main landmark in Brazil’s institutionalization of public policies for women was the creation in 2003 of the National Secretariat for Women Policies (Secretaria Nacional de Políticas para Mulheres—SPM), an organization that had ministry status and was linked to the Presidency of the Republic (Pitanguy, 2003; Calazáns & Cortés, 2011). Another important milestone was the creation of the National Pact to Combat Violence against Women (Pacto Nacional pelo Enfrentamento à Violência contra as Mulheres—PNEVCM), in 2007.

Although these milestones were adopted at the federal level and are valid throughout the national territory and the center has promoted pro-women rights by enacting several national laws as Fig. 1 illustrates, experts highlight the important role municipalities play in combating violence against women (Prado & Sanematsu, 2017; Nunes, 2017). Moreover, according to the SPM, municipalities have full competence and autonomy to create and maintain specialized services targeting women, such as reference and shelter centers, and the creation and implementation of local campaigns (Brasil, 2011). The PNEVCM, for its part, foresees as one of the municipal responsibilities delivering services for women in situations of violence (Brazil, 2011). Thus, it can be said that Brazilian municipalities have been encouraged to create several pro-women institutions since 2003, with the creation of the SPM.

Brazilian municipalities have considerable autonomy in adopting, creating, maintaining and terminating as many of the following three types of pro-women institutions: (a) pro-women rights institutions (e.g., municipal secretary for women’s rights and a city council for women’s right); (b) pro-women justice institutions (e.g., women’s police station, special court for violence against women, and women’s public defender office)Footnote 1; (c) and women-supporting institutions (e.g., a reference center for women and a shelter center for women facing violence). Municipal autonomy in establishing pro-women institutions is reflected in Figs. 2 and 3, which depict considerable municipal variation in the adoption and creation of pro-women institutions across time in the State of Rio Grande do Sul—our case study.

Fig. 2
figure 2

Source: Munic data base (IBGE 2017)

Evolution of the creation of pro-women institutions in the municipalities of Rio Grande do Sul (Brazil), 2013 and 2017.

Fig. 3
figure 3

Number of municipalities with and without pro-women institutions in 2013 and 2017. Source: Munic data base (IBGE 2017)

Variable operationalization

Our dependent variable consists of five official reports of domestic and general violence against women: (a) serious threats, (b) personal injury, (c) rape, (d) attempted murder, and (e) murder. The records of violence are registered originally in the municipal civil police stations and sent monthly to the Secretariat of Public Security (SPP) of the RS, which analyzes, aggregates and makes the data available for public consultation. The violence against women data derive from the SPP/RS and cover two years, 2013 and 2018. Murder data are automatically registered by the police. Data on the other four aggressions may underestimate the actual cases because filing a report depends on women’s willingness to do so. See descriptive statistics in Table 2.

Table 2 Descriptive statistics

The study assesses the violence reporting effects of seven local pro-women institutions: two pro-women rights, (i) municipal secretary for women’s right, and (ii) a city council for women’s rights; three justice institutions, (iii) women’s police station, (iv) special court of violence against women, and (v) women’s public defender office; and two supporting institutions, (vi) a reference center for women and (vii) a shelter for women facing violence. All variables are dichotomous. Data were collected from the Basic Municipal Information Search (Munic) database, organized by IBGE, the Brazilian agency responsible for conducting the main official surveys and providing diverse information for all 5665 Brazilian municipalities. Pro-women institutions data at the municipal level are collected every four years, limiting the analysis to only two years, 2013 and 2017. Because violence reporting effects of pro-women institutions should be lagged, the 2014 violence reports are matched with 2013 pro-women institutions data, and the 2018 violence reports are aligned with 2017 pro-women institutions. To test the additive effect of pro-women institutions, a continuous variable represents the sum of municipal pro-women institutions, ranging from 0 to 7.

The analysis controls for local economic, political, and socio-demographic factors. Specifically, the study includes (a) gross domestic product per capita (GDP/capita) and reported in thousands of Brazilian Reais, (b) percentage of employed women in the population; and (c) percentage of female population with higher education. Unavailability of yearly data at the municipal level is a limitation. Therefore, data for the two periods (2013/2014 and 2017/2018) come from the 2010 Brazilian census (IBGE). These data should not considerably vary across time because cross-municipal differences tend to be maintained geographically over time, as federal policies affect all municipalities. Moreover, the study explores differences across municipalities and not within municipalities across time.

The analysis controls for four political variables: (a) woman mayors; (b) mayors’ party ideology; (c) percentage of women legislators, and (d) women as police chiefs. Given the male-built and male-defined policies and practices of governmental institutions, violence against women is unlikely to be punished or stopped (Yodanis, 2004) because such violence may be subtly or overtly condoned and encouraged (Dobash & Dobash, 1979; MacKinnon, 1979; Walby, 1990). Therefore, gender representation in key institutional settings should influence women’s decisions to come forward and denounce acts of violence against them, as women will feel motivated and empowered by being identified either with the organizational leader’s gender or with the female representation in the institution. In sum, municipalities with (i) a female mayor, (ii) higher percentage of female councilors, and (ii) a female police chief are expected to exhibit higher number of reports of violence against women.

Data on mayors’ and legislators’ gender come from the RS’ electoral justice agency (TRE-RS, 2019). The dichotomous woman variable receives 1 if the mayor is a woman; otherwise, “0.” The percentage of women legislators is a continuous variable, ranging from 0 to 55.5%. Given the Brazilian multiparty system, with a total of 33 political parties (TSE, 2020), categorizing all parties into the left–right continuum becomes difficult. Instead, the study controls for mayors affiliated with conservative parties, which is easier to categorize. We follow the typology proposed by Maciel, Alarcon & Gimenes, (2017), in which 14 parties are aligned to the right wing. To control for police chief gender, data were collected through historical internet research since these data were unavailable due to security concerns. Although the authors were able to obtain a list of women civil police chiefs (or delegate) for the state of RS, this list fails to specify the municipality of workplace. Consequently, the authors conducted internet searches for all female police delegates to identify their assigned municipal police stations.

The analysis also controls for several socio-demographic factors, such as, (a) total population; (b) percentage of Afro-descendant population; (c) percentage of evangelical population from Pentecostal religions; and (d) percentage of rural population dedicated to the agricultural sector. Data for these variables derive from the 2010 Brazilian census (IBGE). In terms of race, the IBGE considers five racial types in Brazil, including white, black, mestizo, yellow (Asian) and Indian. We consider as Afro-descendant population only the population classified by the IBGE as ‘black.’ Another important observation is that IBGE’s classification for religion distinguishes only Catholics and evangelicals. Unlike Catholics, evangelical population groups have several denominations, with Pentecostal churches being very predominant.

Finally, the study also controls for the number of human rights NGOs per thousand of inhabitants. The Private Foundations and Nonprofit Associations, Fasfil database and IBGE, provided the data on NGOs and the values correspond to 2013 and 2016. This is a continuous variable capturing the number of non-governmental organizations that act directly in the defense of human and/or social rights, including support for women at risk.

Data analysis

Recall we have cross-municipal data for 496 municipalities for two periods, 2013/2014 and 2017/2018, resulting in a balanced panel data set with 992 observations. Given we are interested in testing the police reporting effects of pro-women institutions across municipalities, along with the nature of our data, we report random-effect estimates for panel data. The number of existing pro-women institutions within a municipality tends not to change over time. However, this number does change considerably across municipalities. In fact, data show small internal municipality variability in the number of pro-women institutions between 2013 and 2017, the years of study.

Moreover, the Hausman test (for fixed versus random effects) and the Breusch–Pagan LM test (for random effects versus OLS) suggest random effects is the most appropriate model. The variance inflation factor (VIF: 1.49) for a model, including all variables, suggests multicollinearity is not an issue. Five indicators capture types of violence against women, which are our dependent variables: threats, injury, rape, attempted murder, and murder. Therefore, a regression model was estimated for each type of violence report (see Table 3 and 4).

Table 3 Explaining Police Reports of Violence against Women (2014 and 2018)
Table 4 Explaining Police Reports of Violence against Women (2014 and 2018)

Results and discussion

Table 3 presents the results of the random-effect estimates for each of the five violence categories. The Rho parameter shows that each model, respectively, predicts 52% of the variation in threat reports, 41% in injury, 7% in rape, 16% in attempted murder, and 0% in murder. Contrary to the expectations, neither pro-rights (e.g., municipal secretary for women’s rights and city council for women’s rights) nor supporting pro-women institutions (e.g., shelters and reference centers) seems to statistically influence police reporting. Therefore, hypotheses 1 and 3 fail to receive support.

Findings show that only justice system pro-women institutions—police stations for women issues and public defenders for women issues—have a statistically positive and significant effect on police reports of violence against women. According to model 2, the coefficient on women police stations is positive and statistically significant at the 1% level for reports on personal injury. Likewise, according to models 1 and 4, the coefficient on women public defender offices is positive and statistically significant at the 5% level for reports of threats and attempted murder. Therefore, hypothesis 3 receives support, although the third justice system pro-women institution—special courts for violence against women cases—fails to be statistically significant.

In Table 4, five models tested the police reporting effects of the total number of municipal pro-women institutions (additive effect). Results show no pro-women institutions additive effect on police reporting for any of the violence categories. Therefore, hypothesis 4 is rejected. Meanwhile, several control variables report a significant relationship with police reporting of violence against women. Specifically, first the percentage of employed women population is negatively correlated with threats, injury, and rape reports. If the decrease in reports is not due to underreporting, this finding may be in line with WHO’s (2002) observation that communities with larger unemployment rates are more likely to experience violence, including violence against women. Women’s access to an individual income should result in greater autonomy and confidence, thereby reducing the barriers that often prevent reporting of suffered violence (Bhattacharya et al., 2009; Anderberg et al., 2013).

Second, municipalities with larger rural populations are negatively correlated with threats, injury and rape reports, suggesting underreporting due to geographic distance and lack of women’s access to pro-women institutions. In fact, few rural areas in Brazil have police stations, and pro-women institutions are mainly located in urban areas (Hetling, 2000). Finally, political variables are negatively correlated with police reporting. Specifically, municipalities led by female mayors are negatively correlated with threats and injury reports. These findings go against previous studies (Meier & Nicholson Crotty, 2006; Andrews & Johnston Miller, 2013), showing that greater women representation in leadership positions encourages women to voice mistreatments and aggressions. Likewise, municipalities led by a mayor affiliated with a right-wing party is negatively correlated with threats and injury reports. Finally, municipalities with higher numbers of pro-human rights NGOs are negatively correlated with threats, injury, and attempted murder reports. These findings call for further research to disentangle whether right-wing governments and pro-human rights NGOs contribute to decrease violence against women, or whether this negative relationship is a by-product of underreporting cases.

In general, findings suggest the pro-women institutions effect on policy reporting of violence against women seem contingent on two things: the type of pro-women institution and the type of aggression. As expected, the existence of justice system pro-women institutions appears to encourage women to report personal injuries, threats, and attempted murder. Battered women may feel safer reporting the aggressors when a municipal police unit specializing in serving women exists. These results are in line with Ellsberg & Emmelin (2014)’s study which reports that the existence of women police stations contributes considerably to the application of laws. The rationale for this positive effect on police reporting may point at the role of women representation. Both women civic police stations and public defender offices for women usually are staffed with women and with police officers trained to deal with the peculiarities of serving this specific audience (Brasil, 2006). Therefore, results may suggest that gender identification boosts women’s trust and confidence leading them to speak out about their ordeals.

The lack of a statistical effect of pro-women right institutions on police reporting may indicate their effect on women’s decisions to report aggression is symbolic rather than substantive. Having a municipal secretary and/or a city council for women’s rights may be a necessary but not a sufficient condition to trigger women’s decisions to report both the aggressor and the aggressive act. As results demonstrate, only justice system pro-women institutions seem to prompt women’s willingness to file a complaint. The same rationale also may explain the lack of effect of supporting institutions (e.g., shelters and reference centers) on police reporting.

Conclusions and limitations

This study contributes to the limited body of empirical research on violence against women by investigating the influence of institutional factors on violence reports. The study innovates by: (a) conducting analyzes that consider different types of violence against women, from serious threat to murder; (b) testing the effect of different types of pro-gender institutions on violence reports; (c) controlling for a large set of contextual variables in the explanation of reports of violence; and (d) using local data from municipalities in Brazil, one of the countries with the highest number of cases of violence against women.

As Brazil has many laws that criminalize different types of violence against women (see Fig. 1), doubts exist about its effectiveness in deterring violence against women and encouraging women to report mistreatments. Do pro-women institutions influence police reporting of violence against women? Results seem to be contingent on the types of reporting case and of pro-women institution. Having a municipal police station for women issues boosts reporting cases of personal injury, suggesting the need to reinforce the role of women police stations (Lievore, 2003; Flood & Pease, 2009). Results also suggest that having a public defender officer for women issues boosts reported cases of threats and attempted murder. These factors are encouraging news in the long search for solutions. Findings show pro-women institutions help channeling of women’s aggression complaints, suggesting that perhaps other pro-minority institutions (e.g., refugees, religious, gender, ethnic, etc.) may alto trigger benefits.

Our research does have some limitations. The main limitation is the potential of underreporting cases of violence against women reports. On one hand, the analyzed indicators of violence against women capture the cases that have occurred. On the other hand, the reporting cases studied also may capture battered women’s willingness, freedom and/or power to report aggressions. Of the five indicators of violence against women, only murders are reported regardless of the victims’ decisions. The other four indicators fail to differentiate between reported and unreported crimes. As mentioned before, four dependent variables cover exclusively reporting cases, leaving out the potential underreporting cases and reflecting women’s willingness to file complaints.

Another limitation deals with the unavailability of municipal yearly data for economic, demographic, and cultural indicators. Moreover, while this study focuses on the large number of municipalities of the Brazilian state of Rio Grande do Sul, it falls short of comparing findings with other states. Therefore, we caution readers against generalizing our findings and call for future research to test the violence reporting effects of pro-women institutions in other states, countries, and levels of governments. Future studies also should explore the conditions under which pro-women institutions are more likely to empower women. Future research should investigate the drivers of the adoption and establishment of pro-women institutions in Brazilian municipalities. Since the findings of this research indicate the existence of some types of pro-women institutions encourage women to report their aggression experiences, it is essential to understand the conditions under which pro-women institutions emerge.