This is the 15th in this year’s series of posts by PhD students on the job market.
Sexual harassment is a ubiquitous phenomenon for women and can have debilitating impact on the psychological, economic and social lives of women. Sexual harassment awareness training has long been advocated by policymakers, lawmakers, and academics as a way to tackle sexual harassment in universities and workplaces. But we know little about whether this training works to reduce actual harassment and change attitudes about what is ‘correct’ behavior, let alone if such training influences other aspects of interactions between men and women.
In my job market paper, I conduct a randomized controlled trial in the context of educational institutions in India to explore these questions. I start with a model of interactions between men and women building on the literature on social image incentives. There are two types of men: those who prefer to take good actions and those who prefer bad actions. Bad actions constitute sexually harassing behaviors. Men also care about what other’s think of their type. Specifically they care about the disapproval that bad-type men get from their peers. Men can take actions that are not aligned with their true preferences. Women decide whether they agree to the action that men take towards them and would prefer to be matched with good type of men. But since women and men do not know the type of a specific man, they use a man’s actions as signal of his type.
The treatment in the experiment is designed to potentially change three things as suggested above. First, I study if the treatment changes a man’s type (that is, increases the share of the good type) and so then sexual harassment would decline. Second, as well, I study if relationships increase since women will learn over time that more men are a good type. Third, if treatment changes what a man believes about his peers’ attitudes towards sexual harassment, specifically that there is greater peer disapproval about sexual harassment, sexual harassment will decline, but there will also be a decline in relationships if women do not think that men’s types changed.
Experiment
I collaborated with the NGO Safecity to provide group training to a random sample of male students in two colleges. New Delhi is a relevant context for this experiment given the high prevalence of sexual harassment. The sexual harassment awareness training was provided by NGO trainers to men in randomly selected classes for a total of three to five hours per training for each class The training had two main components: awareness and empathy building. The first component provided men with information on sexual harassment and laws against it. The empathy-building component focused on explaining why sexual harassment was a negative behavior.
For the study, I surveyed both female students as well as the male students in the treatment and control classes (1,248 and 1,838 women and men respectively). The majority of students have parents who have completed more than secondary school of education, belong to the low castes (61%; low castes are historically disadvantaged communities in India), and were from Delhi (61%). Only 23% had a working mother. Of the baseline sample, 82% were re-surveyed approximately 3 months later.
Sexual harassment data was collected from women in classes of treated and untreated men, both prior to the training and 3 months after it. I undertook various steps to minimize bias from under-reporting, stigma, selection into the sample, under-detection and privacy issues discussed in detail in the paper. Finally, to study interpersonal relationships, in addition to survey information from men and women, I conducted a lab-in-the-field experiment to understand changes in preferences to interact with the opposite gender.
Does sexual harassment training work?
Yes, it does. The training results in a significant fall in sexual harassment by 0.06 sd. In addition, extreme forms of sexual harassment decline by 1.1 p.p. for female students in the classes of male peers who had the training. This translates to 51 fewer women out of 1,200 experiencing extreme forms of sexual harassment during one academic year. See the table below. This result is not driven by differential reporting or differential awareness of sexual harassment by women (details in the paper).
Why did sexual harassment decline? I find that men’s own intrinsic attitudes did not change but their perception of peer disapproval changed significantly. Intrinsic attitudes were measured along two dimensions: victim blaming (a list experiment to test whether they blamed women for sexual harassment) and support for anti-sexual harassment NGOs (using administrative data).
The implications of this training are broader than the incidence of sexual harassment. Women in treatment classes correctly did not report a change in their perception of the intrinsic attitudes of men in their class. But the training did impact relationships between men and women, especially their romantic relationships, with a decline of 1.3 p.p (which is 64% of a base rate of 2% in the control group). That is, the training results in a higher degree of social segregation by sex. In a lab-in-the-field experiment, I find that this is driven by women, since they preferred to cooperate with women rather than men in their classes on an experimental task. I did not find any such change in men’s choices because of the treatment. This is consistent with the mechanism that the training changed some men’s behavior, not because they think harassment is the wrong thing to do, but rather because they increasing think it will be disapproved of by their peers. If women know this, then they react by being more cautious towards men.
I also find that the results on the decline in relationships between men and women were driven by women in their first year of college. This is consistent with asymmetric information about men’s types. Women in their first year of college are likely to have less information about the type of men in their classes than older students, and hence might rely more on behavior of the men to infer their type. But with some bad type men hiding their type using good actions, men’s actions are no longer as informative thereby making women more cautious.
Policy implications
Sexual harassment awareness training is advocated by lawmakers and academics and is mandatory in many contexts or institutions. But does this training change outcomes, and, if so, what is the mechanism of this effect? Using a low-cost and arguably easy-to-scale intervention, I find that it does reduce sexual harassment (and extreme forms of sexual harassment) and that a key mechanism behind this is what men think about their peer’s attitudes after such training. This might, however, reduce inter-personal relationships between men and women which can have implications for other economic outcomes. Networks formed in such environments can provide future job referrals and have other benefits that might get lost. But if women are able to avoid interactions with bad type men, then it has welfare gains even with reduced relationships.
Karmini Sharma is a PhD student at the University of Warwick.
Join the Conversation