Karmini Sharma on Sexual Harassment, Empathy Training and Gender Bias

Shruti Rajagopalan speaks with Karmini Sharma about the effects of training about sexual harassment and empathy on male and female students

This episode is the sixth in a miniseries of weekly short episodes featuring young scholars entering the academic job market who discuss their latest research. In this episode, Shruti talks with Karmini Sharma about her job market paper, “Tackling Sexual Harassment: Experimental Evidence from India.” They discuss how training about sexual harassment affects men’s behavior and women’s beliefs, how long these effects are likely to persist, and the broader implications for the #MeToo movement in India. Sharma is a Ph.D. candidate in economics at the University of Warwick. Her research focuses on the intersection of economics of gender, development economics and experimental economics. She seeks to understand deterrence of sexual harassment, gender segregation and discrimination.

SHRUTI RAJAGOPALAN: Welcome to Ideas of India, a podcast where we examine academic ideas that can propel India forward. My name is Shruti Rajagopalan, and this is the 2021 job market series, where I speak with young scholars entering the academic job market about their latest research on India. I spoke with Karmini Sharma, a Ph.D. candidate in economics at the University of Warwick. She has a B.A. in economics from Delhi University and an M.A. in economics from the Delhi School of Economics.

We discussed her job market paper titled “Tackling Sexual Harassment: Experimental Evidence from India.” We talked about the impact of sexual harassment and empathy training on men’s behavior toward women, its effect on women’s beliefs about men, its impact on networks and selection, the implications for the broader #MeToo movement and much more.

For a full transcript of this conversation, including helpful links of all the references mentioned, click the link in the show notes or visit mercatus.org/podcasts.

Hi, Karmini. Thank you for coming on the show. It’s great to have you.

KARMINI SHARMA: Hi, Shruti. Thank you for having me as well. I’m super excited.

Effects of Sexual Harassment Training

RAJAGOPALAN: Great. You have conducted a very interesting randomized control trial that you ran at two colleges in Delhi University. At one of them, of course, you give the men an intervention in the form of sexual harassment and empathy training, and in another, you have the same intervention for both men and women.

What you find is both surprising and unsurprising, right? You find that there’s a very significant and clear overall reduction in sexual harassment within the treatment group, but only for the men who received the treatment, which is the people within the college. The women’s experiences with men outside the college doesn’t quite change. That’s a remarkable result. Can you tell me more about this paper? How you designed this experiment? What made you think of it? What were the challenges and so on?

SHARMA: I think the motivation for the paper, in general, was—I come from Delhi, and we know sexual harassment is a severe problem there. This was a problem that was there in my mind for a very long time, and my Ph.D. gave me the motivation to just go after it. I saw a number of papers that had come up at that time, and #MeToo was at its peak.

That was the background of where this project came. Then the call for involving men, I think, turned out to be an important aspect of the project. What I did was I approached colleges in New Delhi, and I designed this experiment where I decided that I will randomize classes in these colleges to receive sexual harassment awareness trainings for men. Why I go with sexual harassment awareness trainings is because these have been really advocated by policymakers, lawmakers around the world, but we do not have very clear evidence of whether these affect sexual harassment or not.

What I do is that when I provide these trainings to treatment classes, I then look at the effect of the treatment on sexual harassment reported by women. That helps me to track the effects of the treatment on men onto actual sexual harassment incidents. As you mentioned, I collaborated with two colleges to do the intervention with just the men, and with one college to do the intervention with just the women.

What I find is that about three to four months after finishing these trainings, sexual harassment reported by women goes down because of the treatment that we had done with the men. And specifically, I can tightly link it to the training because these women tell me that they were less likely to be sexually harassed by men in their own class. That was the unit of the treatment of the experiment. That’s one thing.

What I further do in the paper is develop a theory to understand where these effects might be coming from. This is a simple theory, and just to simplify our understanding of this world, you can assume that men can decide whether they will choose certain actions towards women. They can take, let’s call them “bad actions” and “good actions.” Bad actions are sexual harassment, and good actions are non-sexually harassing behaviors.

Now, the model is set up as such that bad-type men, they would prefer to do the bad actions. The good-type men would prefer to do the good actions; they do not want to sexually harass women. Now, what I’m showing in the paper is that what the treatment ends up doing—or the results are consistent with the theory that these treatments can end up increasing incentives for the bad-type men to pool with the good-type men on good behaviors.

That is happening because they want to avoid social image costs from their peers, which I’ll speak about in more detail. But the key point here is that this increased pooling or increased mimicking, we may call, creates a screening problem for women. Because now, women are in a fix that there are so many men who are acting well, but who are the good types and the bad types now? I show that this has an implication on relationships between men and women, which is crucial in this setup, and I find that that goes down.

What Motivates Changes in Men’s Behavior?

RAJAGOPALAN: That, I found actually fascinating about your paper. What you find is not just that the harassment goes down, but there’s also a decline, let’s say, in the number of romantic relationships within the classroom. This is a fairly common thing that you would find in Delhi University colleges, so it’s an interesting result.

I have a question here, a broader question about preferences. You say that the women’s preferences have changed because there is this pooling or mimicking behavior, and the bad men are signaling in the same way as the good men.

Why don’t you see a change in the preferences of the men, too? Or is it just that it’s not observable? Or you didn’t test for it in survey or something like that?

SHARMA: I’m not seeing women’s preferences, per se, change. Women’s beliefs change. It’s not as if women are becoming “men-haters.” They still want to be in relationships with men. But because of the treatment and because of the pooling behavior that I’m proposing, women’s beliefs are changing about men, and that reduces relationships.

I think one of the most important points you made is, why is it not for the men? I think the important point regarding men’s beliefs or men’s behavior . . . precisely for that, I had designed a lab-in-the-field experiment for the paper wherein what I did was I paired men and women—well, the students together with their classmates in this experiment.

They were randomized either into opposite-sex or mixed-sex pairs, or they were randomized into same-sex pairs. In that experiment, I asked these pairs to perform experimental tasks together, and I asked them privately and separately, “Do you want to do it with your partner who’s been assigned to you, or do you want to do it separately?” Now, there, what I find is that women are more likely to want to do it with other women rather than men in their classes when the treatment’s been provided. On the other hand, for men, I’m not finding such effects.

That was also against my intuition because we would expect that providing these trainings to men makes them more cautious. That’s where it helped me to think through the framework of what might be happening, and that’s why I’m a bit more confident in stating that my results are more consistent with women’s beliefs changing rather than men changing their intrinsic preferences in this case.

RAJAGOPALAN: Well, I think that lab experiment, in addition, is really interesting. When I read the paper, I saw how carefully you documented, anonymously, that women signal their intention to want to work with other women as opposed to men. Coming back to this question of, can men change their beliefs or preferences, I have a question in two parts.

One is, how long do you think the effect of this sexual harassment sensitivity and empathy training lasts? Because you went back after a few months of the training to collect the results. The second is, do you think this kind of training can genuinely change the “bad men” into being less bad or even good?

SHARMA: Yes, I think the first part to your question—one of the mechanisms in the paper that I’m proposing is that these treatments can change men’s perception of, now, what do they think their peers are going to do after these trainings. What I call this is “perception of social disapproval.”

If men start seeing more social disapproval or perceiving more social disapproval, this can give rise to this pooling behavior that I was talking to you about earlier. This is not to say that some of the men might not have intrinsically changed or that awareness did not affect men’s behavior. But it seems that, in this case, the first mechanism is stronger than the other two mechanisms that we are talking about.

If I think about just these college students, because they are in these classes for quite a long time together—after this college, their peer groups are going to change. If this mechanism is more important and it’s stronger, then I would say that it possibly could have a different effect on men’s behavior. Maybe they’ll revert back to the same old behaviors if that was a key constraint. Unfortunately, I cannot test that.

RAJAGOPALAN: Actually, another way of testing that—not just in the future, but in the present—is if the men will receive the treatment, if their behavior changes outside the college setting, right?


RAJAGOPALAN: With women who are in the classroom, the cost of bad behavior has now suddenly increased, so as one would expect, you see less of it. But if the cost of bad behavior has not changed outside the classroom setting—let’s say at the canteen or on the bus ride, or on the metro to college—then you would see more harassment there. I can understand your constraint of not being able to test for that and not being able to observe every interaction they have with women outside of the classroom.

SHARMA: Exactly, that’s absolutely that. That’s why I’m confidently able to say that these men’s behavior has changed within the classes, because I’m collecting these sexual harassment reports from women instead of these men. Outside the college—exactly, as you said—tracking their interaction with other women, collecting data from those women is, of course, impossible. But also the idea that collecting the data directly from men—that’s even more difficult and less credible.

Effects on Women’s Experience Outside the Classroom

RAJAGOPALAN: In questions of interventions like this, one of the normative implications of this kind of intervention is that it’s going to change women’s choice set. As we know, especially in a place like Delhi where about 93% of women report sexual harassment in the workplace and public spaces, on the way to the workplace or college, and so on, things are quite difficult, and it dramatically changes their everyday life.

For instance, Girija Borker’s work shows that women change the travel route that they take to work. They actually almost spend twice as much time traveling and way more money to have a safer way to get to university, relative to men, so that they can be protected from sexual harassment.

Now, those are really important consequences. In some sense, these kinds of interventions have these big normative implications. Having said that, when it is done in such a specific and localized way, do you think there is any big difference in the everyday lives of these women, given that the classroom interaction is so small and their larger life choices haven’t really changed in a major way?

SHARMA: I think even though I’m doing this in a localized setup, colleges in New Delhi, the thing is that the environment or the setup that we are dealing with here is what I call “interpersonal settings,” and they have repeated interactions between these men and women.

This is very similar to other settings as well. For instance, workplaces. I think my results directly speak to those settings as well. With regard to actual choices of these women, I guess the fact that we are seeing this decline in sexual harassment, although I don’t track them, this will have its direct welfare effect on these women.

For instance, the translation of the effects is around 50 women are saved from experiencing severe forms of sexual harassment over six months, which has a huge economic benefit. As far as results on relationships are concerned, however, the fact that these women are choosing to interact maybe less or date these men less—the point is one thing I cannot say, whether these men are the abusive kinds with whom they are changing their relationships.

There is that benefit that we might not be able to see. Having said that, yes, it will change network structure. That’s what I’m showing in the paper, that there is segregation, clearly, and that will have repercussions for how networks develop for these men and women over the future. But exactly the welfare effects of whether this was good or bad for the women? It’s still an open question in that sense.

Male-Female Interactions Inside and Outside the Classroom

RAJAGOPALAN: Absolutely, I can understand that, and it’s also very soon. I want to dig in a little bit into the segregation part. My intuition is that if the segregation is because of this pooling and mimicking that is taking place within the classroom, then the women report fewer romantic and other interactions with men within the classroom, but they should report the same number of interactions with men outside the classroom—or maybe even more, if there’s a substitution effect, right?

Do you see any of that taking place? Because they can’t tell the good apples from the bad apples within the classroom, are they looking outside the classroom or outside the college for romantic engagements?

SHARMA: Exactly, that’s exactly what I find. I find, in fact, that women’s relationships with men outside the class increases. That’s evidence that women are actually actively substituting away from treated men towards men outside the class who’ve not been treated. I mean, I’m happy about this result, to be honest, because it shows that women’s overall utility in matching with men does not go down, so “not becoming man-haters,” but it’s really the treatment that is affecting women’s choices here.

RAJAGOPALAN: This intervention through this sexual harassment training—do you think it has made the classroom environment a little bit more awkward, and that’s the reason for this decline? It’s just people really don’t know how to talk to each other anymore, and your usual rules of the game, which are quite fluid, have now become a little bit more standardized because you’ve gone through this training, and it’s going to take time for both genders to seriously find a way to navigate this?

SHARMA: I think if that were the case, for instance, what I would find is perhaps that men are approaching women much less because now men are in this awkward situation where they don’t know what to do, how to navigate this. However, I don’t find evidence for this.

It tells me that it’s not as if men have changed their relationship offers in a way—not just relationship offers, but them approaching women to interact with them. That has not changed. At least, I don’t find that in the data. I have not collected direct outcomes on how awkward interactions became, but it is possible. It is still possible.

RAJAGOPALAN: Honestly, it’s a great result that harassment is down. That’s not what I’m questioning. I’m really just trying to think through, how long do these effects last? That’s what I’m trying to get at from different points of view. Does the training start going away? Do women get better over time at separating the good apples from the bad apples? Do people overcome their awkwardness, if any, post a very rigorous sexual harassment training program, and so on?

That’s what I’m very curious about coming from your paper. Do you think you’ll be in a position to go back and look for results later in time? Or is it problematic because everyone just disperses year after year in college?

SHARMA: I think two things to that, because, in general, the larger concern that you have about how long these results will last or what—yes, I guess that’s the main point. I find that the results in opposite-sex relationships, where I find that women are reducing their interactions, that’s much stronger for first-year students than it is for second- or third-year students.

One of the reasons behind that is because these women do not know enough about these men. There is this lack of information about these men as well, because they’ve not interacted with them before. One good implication out of it, if I may say, is that once they get to know them more, I’m hoping that this kind of screening will become easier and, hopefully, interactions will become better.

Principals, Agents and Gender Bias

RAJAGOPALAN: Can we take a bit of a detour from your work in the field and go into your work in the lab, where you’re trying to understand gender bias when it comes to principals evaluating the performance of their agents? This is, of course, your paper with Fenske and Castagnetti.

SHARMA: Yes, I’m happy to talk about this. This was a lab experiment. In this paper, we are trying to understand or study the role or the presence of attribution bias by gender. The key hypothesis, in this case, is that in many settings, outcomes from work are a function of luck, factors that are outside the person’s control, and also that person’s own ability and effort.

One of the things that we wanted to test is that, is there a differential tendency of, say, bosses, or principals in this case, to assign successes or failures on tasks to ability and luck for men and women differently? One, for instance, manifestation of this would be that a male or female manager, when they see a female employee lose on a particular task or fail on a particular task, are they more likely to say that that’s because of her ability versus a male employee?

And if, suppose, she does win or succeed on that particular task, are they more likely to say it’s because of her luck or factors that were in favor of her, but not just her ability more than the men? Thankfully, we do not find any result; we do not find the presence of attribution bias. Yes, lab is an interesting setup. [laughs]

RAJAGOPALAN: Do you think the lab is a very specific context where the typical cultural and social context in which this kind of discrimination can take place gets completely removed, in one sense? It’s a great way to study the problem, but do you expect that you wouldn’t find these results outside the lab given the nature of the question?

SHARMA: That’s a very interesting question. One of the things that we are, in fact, exactly trying to do to address a question like that—that is, to improve external validity, in a way—is that we are planning on expanding the sample, or in the process of expanding the sample, to other countries. This one was done in India, and expanding it to other developing countries and developed countries, that’s when we’ll be able to say, “Is culture a factor?” Yes, that’s exactly what we are going to do.

Implications for the #MeToo Movement

RAJAGOPALAN: I want to maybe to step away from the precise research papers we’ve been talking about. I want to talk to you about the larger question of #MeToo. You told me before, that’s how you got into this research area. Of course, you and I are Delhi girls, and we’ve experienced this everyday minor and major sexual harassment for most of our lives.

I want to think through, what do you think are some of the implications of your research on the #MeToo movement, specifically in India, in large urban centers?

SHARMA: That’s a big question.

RAJAGOPALAN: Let me break it down in a few parts, or let me share what I was thinking through when I read your paper, and you tell me if you agree or disagree, maybe, if that works? My sense from reading your paper—there’s, of course, one result that should please both of us, which is the number of instances of sexual harassment that have been experienced for these 45, 50-odd women have reduced, and so the treatment is working. That’s very heartwarming.

On the other hand, if your mechanism—where you describe the mimicking and, therefore, the pooling effect where the bad apples are mimicking the good behavior of the good apples, and the women can’t tell the difference between the two. If that mechanism holds, then is it a cause for concern that we live in a world, in India, where the law enforcement system is completely broken? #MeToo cases either don’t get a hearing, or they get dragged on for years and years. Or in some instances, it’s actually the person who is the whistleblower who is getting harassed, as we saw in the case of Priya Ramani and M. J. Akbar.

The question I’m really asking is, given that our law enforcement system is completely broken, do you worry that this kind of sexual harassment intervention and training is going to help bad apples hide in plain sight—given that women don’t have much of a recourse, other than training programs, to bring down any form of formal complaints or report it to the police and actually get some form of justice through law enforcement?

SHARMA: Again, I think a super important question, and interesting. The results that I have are, again, for three to four months after the intervention. And it is possible, over long run, there is a change in intrinsic attitudes of these men, and maybe it’s difficult to change these attitudes over a shorter run. Of course, the thing that makes me a bit more pessimistic is that these are young adults, and so their attitudes might be a little less malleable than schoolchildren, for instance. That has been shown in other literature.

Now, in terms of these bad apples hiding—one of the things that I said about heterogeneity of the results, that if it is the case that more interactions can help remove these information constraints that women might have about types of men, then it is possible that it’s not possible for men to hide so easily if there is enough data that women have from interacting with these men. But what we find for the first-years, where there’s not enough information, clearly there is that implication that some of these men might be just able to change their behavior, even if they don’t change their intrinsic attitudes.

For me, I will still take that than them not changing their behavior and continuing to sexually harass women.

RAJAGOPALAN: Really, good behavior is on the margin making the lives of these women much, much better, and it’s overall reducing the culture of harassment. I think that’s a very strong result. That’s something to be quite pleased about. Do you think there are some other implications for the #MeToo movement? Any other questions you’d like to study on the larger topic?

SHARMA: I guess a lot of the emphasis or a lot of the dialogue has been about what women should do, whether women should complain or not. “Why didn’t she come forward before? Why didn’t she complain before?” and all of that. I guess it has to very consciously pivot towards the men. It has to consciously pivot towards the norms, in general, or what the institutions are like, and what are they doing to ensure that this does not even happen. Because now we understand very well that it’s difficult for women to talk about this, and they are traumatized when they complain.

The way that I perhaps want to contribute is that it is possible to engage men, it is possible to have a dialogue with men, and it is possible for it to have an effect on sexual harassment that we care about and that removes the burden from women. That’s the main thing, I think.

RAJAGOPALAN: No, I think you’re absolutely right. I think a lot of your research is showing how changing the relative cost of behavior—or observable behavior, at least—in the classroom has this big impact. We need to form stronger norms and a stronger culture where you can continue to keep these costs on men, and not just on the women as we’ve seen so far. What is the big research program going forward?

SHARMA: In my future work, I want to extend my work to the workplace, as I alluded to, because the environment of these universities is, again, very similar to the workplace. And so, there, of course, there are clear repercussions for productivity, absenteeism and things that we directly care about as economists as well. That’s where I want to move the agenda, and hopefully you’ll hear more.

RAJAGOPALAN: Oh, yes, absolutely. I think you have lots of work left to do, and given that these are such specific and contextualized results, you’ll have to replicate these in many, many places. So I imagine you’re going to have a very busy next few years.

Pandemic Activities

RAJAGOPALAN: A couple of other questions, just to get to know you a little bit, Karmini?


RAJAGOPALAN: How have you been coping with the pandemic?

SHARMA: [laughs] Cooking a lot. I really changed my diet during the pandemic; I don’t know why. And I started focusing a lot on healthy cooking, and so I started following a lot of YouTube content creators.

RAJAGOPALAN: Oh, that’s great. I should ask you for some recommendations if you have any.

SHARMA: I can certainly send you the links to these content creators. I don’t know if you know the concept of Buddha bowls? These are very healthy. I don’t remember their names, but I follow a lot of them. I’ll send you links.

RAJAGOPALAN: Now, for the most important question: What have you been binge-watching?

SHARMA: Yes. “Queen’s Gambit,” “The Crown,” and also, as I said, I’m not particularly proud of it, but “Keeping Up With the Kardashians.”


RAJAGOPALAN: Hey, you got to do what you got to do. Thank you so much for making time to chat with me, Karmini. It was a pleasure reading your papers and speaking with you.

SHARMA: Thank you much. I’m truly honored and grateful to get a platform like this. Just hearing your past podcasts as well, the work that you’re highlighting, thank you. Thank you for that.

RAJAGOPALAN: You’re most welcome, and we hope you’ll come back in the future and share more of the results.

SHARMA: I am tracking these men and women. Now it’s been around more than a year since the interventions. Hopefully, we will have the dataset, the end line completed soon, and I’ll have those long-term tracking results for us to discuss.

RAJAGOPALAN: Oh, fantastic. Then I will have to call you back to talk more about it.


About Ideas of India

Host Shruti Rajagopalan examines the academic ideas that can propel India forward. Subscribe in your favorite podcast app