In today’s Q&A with Members post, we feature EGAP member Sabrina Karim (Cornell University) who spoke to us about her work on the role of gender inclusion and gender norms on police performance and citizen trust in security institutions.
Sabrina Karim is the Director of the Gender and Security Sector Lab. Launched in 2018, the Gender and Security Sector Lab is partnered with the Geneva Centre for Security Governance (DACF), and UN Women’s Elsie Initiative Trust Fund and aims to understand how gender equality within the security forces affects police personnel and soldier’s perceptions and level of restraint on the use of force and misconduct. To do so, the lab, which includes five graduate fellows, is conducting surveys of police and military personnel all over the world including in Uruguay, Senegal, Ghana, Zambia, Bangladesh, Liberia, Sierra Leone, Jordan, Norway, and several other countries. The original surveys include survey experiments that prime gender identities, as well as conjoint analysis to better understand discrimination against women in the security forces. Outcomes of interest include perceptions of the excessive use of force, protection of civilians, beliefs about the appropriateness of misconduct, and the security force’s role in politics.
Much of your work considers the role that gender plays in how state agents provide security. How did you come to study this topic?
Sabrina Karim: In 2010, I received a Fulbright Fellowship to conduct research in Lima, Peru. One of the first things I noticed was that all the police officers on the street were women. I learned that the traffic and transit division of the government had feminized the Peruvian National Police as a corruption reduction strategy. This policy change led me to become very interested in the instrumental role women play in the provision of security and governance. This research interest expanded during my PhD, when I started to research how the inclusion of women into peacekeeping operations affects the effectiveness of peacekeeping missions. This research, however, led me to discover that changes in outcomes (e.g, corruption, restraint) are not necessarily affected by women’s inclusion, but rather by gender. Beliefs about masculine gender norms being privileged or rigid beliefs about gender roles could play a more important role than the the sex of police and military personnel if we are concerned about changing the organizational culture of the security forces. Thus, my research differentiates the inclusion of women from gender when trying to develop interventions to change perceptual and behavioral outcomes of the security forces.
Your research has examined how the incorporation of women into police forces affects both the workings of the police and citizen-police relations. Let’s focus first on the inner workings of the police. Based on your research, how does the inclusion of women in the police force impact police operations? How did you investigate this question and what were the most surprising findings from this research?
SB: One of the main policy recommendations by the UN, particularly as a part of the Women, Peace, and Security Agenda, to improve security is to include women in the security forces. As such, countries such as Liberia, which benefited from a UN peacekeeping mission, often see quotas for their security forces. In the case of Liberia, there was a 30 percent quota to include women in the police force. In this paper with Michael Gilligan, Rob Blair, and Kyle Beardsley, we wanted to understand how these policies affect intra-police dynamics including group cohesion and effectiveness. We partnered with the Liberian National Police to randomize the group composition by sex. Each group of six either included 0, 2, 4, or 6 women. We then engaged in a series of activities that measured outcomes of interest such as deliberation, teamwork, participation, discrimination, and decision-making. We found that adding women can increase unit cohesion, but also that adding more women does not lead to an increased awareness of sexual and gender based violence, nor does it reduce discrimination.
Most of the literature suggests that female police officers should be better at handling sexual and gender-based violence (SGBV), but we were surprised to find that the female police officers in our study were not more likely than men to believe that a crime might include SGBV. In other words, women were not more sensitive to thinking about gendered crimes than men. What mattered the most for this outcome was the level of professionalization of the police officer. Additionally, some policymakers are concerned about integrating women into all-male units, such as combat units, because they fear it may disrupt unit cohesion. Our research, however, did not find any evidence that unit cohesion is disrupted when women are included. In fact, unit cohesion can increase with the inclusion of women.
Turning to citizen-police relations: What do you find in terms of the effect of the inclusion of more women in the police force on citizens’ perception of the police?
SB: In addition to assumptions about women being more effective in handling SGBV, some of the literature and some policymakers argue that women’s inclusion in the security sector is important because it enhances the legitimacy of the police force. This descriptive representation could make civilian women who interact with the security forces more trusting of the police precisely because they are interacting with other women. I first tested this idea through surveys with community members living in informal settlements, a group I thought who had less trust in the police, by asking them about their interaction with male or female police officers. I assumed “as-if” randomization of interactions with male or female police based on the availability of police officers in police depots in Monrovia, Liberia after observing that there is minimal selection bias involved in the sex of the police officers attending to citizens at the depots. The paper finds that interactions with women can improve perceptions of the police in informal settlements in Monrovia, Liberia.
However, given that the above study did not randomize police visits directly, in 2015, I conducted a study that randomized household visits by male and female police officers randomly selected villages in rural Liberia. This field experiment shows that while household visits greatly improved rural residents’ perceptions of the police, there was no difference based on officer sex. I suggest that the type of visit is important for enhancing perceptions—visits that build relationships and that focus on the needs and questions of the residents are likely to have this effect.
The findings suggest that there might be certain conditions under which female police officers can improve trust and legitimacy in the police (e.g. perhaps in certain communities), but they may not enhance legitimacy more so than male police officers. What seems to matter more than officers’ sex for police legitimacy is how police officers—male or female—conduct themselves with the public, particularly when they pre-emptively engage with the community in a positive way.
In some of your research you note situations when citizens appear to place greater importance on the “police officer” identity of agents rather than their sex. How do you explain this result and what are implications for future research in this field?
SB: As mentioned above, this paper randomized household visits by police officers’ sex finds that police visits by both men and women improve perceptions of the police, but that male and female police officers were equally likely to cause this effect. Why might this be the case? I argue that police officer identity might be more important or recognizable than the officer’s sex.
To make sure that police officers’ personalities and/or gendered identities were not driving the results, I chose four police officers—of the same ethnicity—who scored the most similar on personality tests. However, we know that policing is a masculine occupation—and therefore those who become police officers often have to assimilate their identity to match this masculine police identity. Women, in particular, often adopt a masculine police identity in order to fit in within the organization. This means that the four officers I chose for the household visits likely had more masculine personalities than feminine ones. So, if the female police officers “behaved like men” and in particular like male police officers, then the null effects make sense. Rural residents saw the female police officers as the same as male police officers. Moreover, the fact that they were all wearing the same uniform likely reinforced this “policeman” identity.
The findings allow us to rule out officer sex as a possible mechanism for enhancing trust. Instead, future studies should shift attention to developing interventions that randomize gendered identities (e.g.randomize feminine police officers versus masculine police officers), and gendered identities that interact with officer sex (e.g. randomize visits by masculine men, feminine men, feminine women, and masculine women). This would give us insight about whether the gendered identities of police officers could enhance trust in the police among different populations.
Most of your research on gender and security has taken place in Liberia, a post-conflict setting where gender balancing of security sectors formed part of state-building reforms in the aftermath of the civil war. How do you think this context impacts your results and do you expect things to work similarly elsewhere?
SB: After spending many months and years building a relationship with the Liberian National Police leadership, they allowed me to work with them on several interventions. This trust between the researcher and a security agency is difficult to achieve, which is why much of the research to date has been with the Liberian National Police. In many ways, Liberia was an ideal case to study the effect of sex and gender on the security sector because reforms related to women’s inclusion and gender were so integral in the rebuilding process of the Liberian National Police. And, it was a new reform to the Liberian security sector, allowing me to test the effects of visible, novel reform.
The problem with a single case study is that the findings may not generalize across other contexts. There are several conditions that might be necessary for the research findings to travel outside of Liberia. First, women’s inclusion in the public sector, including the security sector, cannot be banned nor can it be the norm. In other words, the inclusion of women in the security sector should be somewhat novel, but should not generate large-scale backlash. In Liberia, rural residents were sometimes surprised to see female police officers and some reported being intimidated at first, but the visits were able to change their minds. Moreover, the findings above would likely not hold for countries where there is a high degree of trust in the police already or where repression by the police is severe and recurrent.
There are limited studies on female inclusion in the security sector, with most of the experimental work being conducted in India and Liberia. Yet, other countries have experimented with gender reforms within the security sector as well, including Peru and Mexico. And, more countries are likely to do so given the proliferation of the Women, Peace, and Security agenda. Thus, there is ample room for more experimental work to be done to better understand how reforms related to women’s inclusion and gender reforms affect citizen’s perceptions of the police, intra-policing dynamics, and police effectiveness, particularly in countries where these reforms are being rolled out.
The Gender and Security Sector Lab is carrying out research, with Laura Huber and Lindsey Pruett, that investigates how individual beliefs about gender roles and toxic masculinity contribute to the use of excessive force by security forces in different countries (Zambia, Ghana, Uruguay, Senegal, and others). What are the main lessons that emerged from this research?
SB: The aim of the Gender and Security Sector Lab is to better understand how gender equality within the security forces affects outcomes related to the use of force and security force misconduct. In this particular paper and future versions of this paper, Laura, Lindsey, and I develop two survey experiments to prime gender as well as develop novel scales of toxic masculinity and rigid gender roles. The surveys have been and are being administered to police and military personnel in over ten countries.
In the first survey experiment, we prime people to think about their gender identity by first asking them to rate their masculinity. We then show them a picture of a hyper-masculine man from their country and ask them about how much they identify with this man. This mimics the many pictures and advertisements of hyper-masculine men that are used for the recruitment of personnel into the security forces. The picture is meant to have them think about their own masculinity as they answer questions about the use of force and misconduct. We find weak evidence that this prime affects perceptions of misconduct.
In the second survey experiment, currently being fielded, we prime gender by first asking people to rank their masculinity or femininity (randomized). We then randomize telling them that they scored above average, average, or below average compared to everyone else who has taken the survey. This treatment is meant to trigger feelings of hyper-masculinity (if they were told they scored above average for masculinity or below average for femininity) or emasculation (if they were told they scored below average for masculinity or above average for femininity). We assess whether this trigger affects their responses to questions about the use of force and misconduct.
Finally, we tested and piloted a number of questions to come up with a toxic masculinity scale and a scale to measure rigid gender norms using principal component analysis and item response modeling. We find strong evidence that these indices are positively correlated with a propensity to believe in the use of excessive force against civilians and escalate a situation into violence as well as a negative correlation with finding misconduct acceptable, reporting it, and punishing personnel for it. The correlations suggest that gender inequality and norms about violence and misconduct are interrelated.
Our future goal is to develop interventions that might shift gender norms within the security forces. If our survey experiments demonstrate that gender identities can be primed through a simple survey, then we gain insight into actual programmatic interventions that could have an effect on mitigating security personnels’ propensity to use violence under certain circumstances.