(with Rachel Brulé and Alyssa Heinze)
in partnership with the South Asia Center & Penn Comparative Politics Workshop
(English captions & Hindi subtitles available)
About the Seminar:
While descriptive representation of women has increased, ensuring their inclusion in political institutions, these institutions rarely include subsequent guarantees to protect the voice of elected women. Indeed, in a survey of 600 Indian villages, Prof. Chauchard (with Rachel Brulé and Alyssa Heinze) demonstrates that women elected through gender quotas do not have equal voice in political deliberation, compared to their male counterparts. Can modifying the institutional rules ameliorate such gender disparities in political voice? They implement an experiment that manipulates the discussion rules in deliberation and measures its impact on the political voice of elected representatives. They find that discussion rules—specifically, rules stipulating that the elected president speak first (or last)—can improve gender inequalities in political voice.
About the Speaker:
Simon Chauchard is a Distinguished Researcher (Investigador Distinguido) at the University Carlos III (Madrid) and the PI of the POLARCHATS ERC project. He previously held positions at Dartmouth College, Columbia University (SIPA), and Leiden University. Much of his current research focuses on the causes and consequences of social media misinformation in developing countries. As part of the POLARCHATS ERC project, his research explores the prevalence of misinformation online, strategies to reduce misinformation or correct it, and the uses that parties make of social media (mainly WhatsApp), in India and beyond. He also co-organizes the DIMIS (Digital Misinformation in the Global South) initiative. In parallel, he continues to produce research on voting behavior, political representation and politicians-citizens relations in India. His work relies on qualitative, quantitative, and experimental methodologies and covers a variety of themes relevant to contemporary Indian politics and beyond, to political processes in the world’s democracies. Current projects focus on preferences for redistribution, and on “proxy politics.” His works have appeared in Political Opinion Quarterly, the American Political Science Review, Comparative Political Studies, the Journal of Politics and Asian Survey, among other research outlets. His book, Why Representation Matters: The Meaning of Ethnic Quotas in Rural India (Cambridge University Press, 2017), combines qualitative work and a series of innovative surveys to explore the impact of caste-based reservation policies on everyday intergroup relations in India’s villages.
FULL TRANSCRIPT:
Naveen Bharathi:
Hello and welcome to CASI's Spring Seminar Series. I am Naveen Bharathi. I'm a Postdoctoral Researcher at CASI along with my colleagues moderate this series. Today's talk is in partnership with the South Asia Center and Penn Comparative Politics Workshop. Before I introduce today's speaker. I just want to put in a plug for our next seminar on March 17th same time by Paul Staniland from University of Chicago, who will be talking about his new book, "Ordering Violence." Please register for this event on CASI's website. Today, I'm delighted to welcome Process Simon Chauchard. Simon is a distinguished researcher at University Carlos Three at Madrid and Principal Investigator of the POLARCHATS ERC project. He has previously held positions at Dartmouth College, Columbia University and Leiden University. Much of his current research focuses on the causes and consequences of social media misinformation in developing countries. His research explores the prevalence of misinformation, online strategies to reduce misinformation and authority and the uses the political parties make of social media.
In parallel, he continues to produce research on voting behavior, political representation and politician-citizen relationship in India. His work relies on qualitative, quantitative and experimental methodologies and covers a variety of themes relating to contemporary Indian politics. His works have appeared in Political Opinion, quarterly, the American Political Science Review, Comparative Political Studies, The Genre of Politics and Nation Survey among other research outlets. His book titled, "Why Representation Matters: The Meaning of Ethnic Quotas in Rural India" was published by Cambridge University Press. In today's seminar, Simon will present his work with Rachel Brule and Alyssa Heinze titled, Why Discussion Rules Matter for Representation, Experimental Evidence from Rural India. April, Simon and his team demonstrated women elected to gender quotas do not have equal voice in political deliberation compared to their male counterparts. Can modifying the institutional rules ameliorate such gender disparities in political voice? They implement an experiment that manipulates the discussion rule in deliberation and measures its impact on political voice of elected representatives.
They find that discussion rules can improve gender inequalities in political voice. Simon will present for 30 minutes and we'll have questions from the audience for around 30 minutes. Please keep your questions brief to the point so that we can get to ask many questions as possible. So if you have any questions at the end, could you please use the chat box to send them directly to me. And I will call on you to pose your question to our presenter. Once again, thank you for your interest for being here today with that I'm going to handle over to mic to Simon. Thank you.
Simon Chauchard:
Thanks Naveen. Hi everyone. Can you see my screen? Yeah. Does it work? Cool. All right. So yes. Thanks Naveen for the introduction. And yes, indeed it's co authored work with Rachel Brule and Alyssa Heinze. They should be around. Maybe you'll get to see them in a Q&A. We're really excited to be presenting, this is the first time we present anything from this project. And so let me jump right in and tell you what this is about. So it's called... This is a project about making political representation matter, right? And as a backdrop, I'd like to start with the sort of facts that of course, as in India an increasing number of members of disadvantaged groups elected in office. And so a large number of them are now access to political office, especially if you're think about local representation. Just if you think about gender, there's now around a 100,000 women who are a Village Council Presidents. In India, sarpanches, pradhans, mukhiyas as here, I'm going to refer to them as sarpanch depending on the region may defer in a little bit.
So this sort of has prompted some politicians to sort of showcase India as a global leader in terms of the political representation of women of course, as of many caveats on this assertion. And I'm sure many of you will disagree for various reasons. We don't especially agree with this, but the fact is local government over the past 30 and especially 20 years has acquired very often a completely new face when you think about gender. Many states now have a 50% representation quota. Here you see a picture of a gram panchayats in Gujarat, that is on all female gram panchayats. It's one of very few, but it exists, right? Now, there are complications, right? And the big one is this, that the presence in office of members of these categories, and here women does not necessarily equate voice their ability to the fact that they speak or that they're influential on decision making once in office, right?
Anecdotally, if you poke around and talk to people around this, you hear a lot references to these elected women labeled as been proxies for diversity of other actors, right? Who might these actors be? Well, they might be husbands of course, but they also by the way, maybe sort of political or bureaucratic actors within the institution itself, right? Sort of colleagues within the village council, wild panches, but also the gram sevak, right? And so what this implies is really that there's a form of political inequality along gender lines after officials reach office. I think in political science, we've studied a lot these gender inequalities in terms of who gets to be a candidate and who gets to be elected, but here we're really interested what happens once people reach office, right? And inequality may be something like that, right? The fact that women are sort of less vocal, less influential, and that they play a less central role in general in the deliberative processes that happen in the village council, right?
So I'm not going to go over all of this, but just so that you could take me sort of 20 minutes to go over this typology, but just so that you know there's a number of different reasons why women may have more difficulty being influential in office, right? Ranging from the fact that there may be sort of political exclusion and straight sort of violence against women in office to much more subtle mechanisms. Some women may be personally disinclined to be in that position, or they may face sort of hurdles or barriers from a variety of factors as I mentioned. But the point is whatever the obstacle is, they are consequential, right? And the consequential fact is two reason. One is of course, that descriptive representation. The presence in office of these women because of these obstacles may not lead to substantive representations, right?
So the fact that these women reach office need not transform into budgets that reflect the preferences of women if these are different, right? Simply because they are not as influential as they should be. And in addition to this, and this is... I will argue actually equally important, the fact that these women don't necessarily have an equivalent voice or equivalent level of influence within institutions may have a chilling effect on broader political engagement by other women, right? So without even thinking about the consequence on budget, there's something that really matters about this in so far as it should have an impact on what other women think and do about politics, right? So in the broad project you have sort of two types of research questions. We're going to talk a little bit about both of these today. One of them is just simply descriptive.
That's really more than one there, but the big one is this how large, or how common is this? Right? We know there's some kind of inequality, but is it everywhere? Is it for every kind of official from a reserved category? That's and how common is it? Right? And the second sort of big thing we're doing in the project is we're trying to think about what can be done to reduce the gender gap and give these newly elected women more of a voice and more influence on the sub deliberation at this level, right? And so what I'm going to focus on today is a really specific part of this, which is an experimental part of the project in which we actually studies the effect of a request that the various actors of the gram panchayats, the request that they all speak during a deliberation. We're studying whether that improves the influence of elected officials from disadvantaged categories. And in that sense, we're studying discussion rules, okay.
I'm stuck here. So just a quick outline just go over the theory for a few minutes, and then research design, some evidence, two types of evidence. Let's see how far we go there, but we have exciting graphs. So if I have to go a little bit fast, I'll send this slide. So theory is just some summar... I'll summarize the argument quickly here while I have your attention, we're simply arguing this, that male and female representative are not equally able to weigh in upon on decision making within deliberative institution. And because of this, women are less able to persuade and influence decision making. However, we also argue that simple tweaks, and I'll give you a concrete example in a second in discussion rules, if they can be enforced, may help reduce this gender gap in influence. So what is our intervention? Right? Our intervention is something it's a really a minor tweak, right?
It's probably qualifies for what we want to call a nudge, right? So the only thing we do is we organize a deliberative process in a sort of controlled environment. And well, we explicitly asks that actors adhere to discussion rules that require sarpanches to speak prior to the beginning of the discussion. So we randomly ask in some locations that people follow discussion rules that require the sarpanch to participate, right? It's a very minor intervention and ready and just to sort of flag something that you'll figure out later. Again, it's a request and you see that it's not completely enforced, right?
The assumption we're making here since our goal is to reduce the gender gap, and that's going to be important when we think about the interpretation of our results is the male officials probably don't need such a nudge to participate, right? And so by providing, by asking, making sort of by clarifying that everyone has to participate, we will close the gender gap by making sure that women have a voice. And so what we do here is of course, the comparison is in between meetings in which this request is made and meetings in which this request is not made. What's the rational for this? I've already told you a little about this, but of course, it's meant to address this fundamental limitation of local deliberative institution, which is that without voice, it's hard to think how women could be able to actually influence the decision of the council, right?
And that's an intuition. So Alyssa Heinze did awesome sort of field work, looking at a ton of meeting in Rural [Maharashtra 00:13:01] before we actually started with the quantitative field work. And one of the big intuition was that this absence of clear and enforced rules for how discussions were to take place was both hindering the voice of women and hence them participation, and as a result their input, right? Our intuition is not sort of purely sort of based on this qualitative insights, though there's also sort of in the literature, a number of important insights about the role of institutional rules in regulating political inequality and outside just one big sort of this inspiration from the project, which is this fantastic book series of experiments as well, by a Karpowitz and Mendelberg in which they do show that institutions and the rules that govern institution actually matter for political inequality along gender lines, right?
In their case, they really focus on the book if you remember on decision rules, right? And they show that changing decision rules can provide decision making processes that are more favorable to women, but they also mentioned, it's a bit of a footnote in the book, this idea that discussion rules also matter. And this is where we opportunistically come in and build on this intuition, right? So it is the voice, guarantees on the voice of elected. This can start a vicious circle that would more or less look like this, right? So if the discussion rules are more inclusive, you'll see more voice and participation. This may happen because the official herself or himself changed their behavior upon the appearance of these new rules of these more aggressive rules, all because others adapt.
And you may think about various mechanisms as to why others would sort of change the way they behave, or for example, interrupt as a result of these rules being now clearly stated. And because they speak more sort of just unfolding here, the sort of causal diagram, because as this they're more able to speak and participate where we have a greater chance, you may want to think of this last arrow as really probabilistic.
We have a greater chance that there are participation results in persuasion that is that they convince other members of the council, other actors present in the deliberation and separately that this actually results in a different outcome, in a different collective preference made as the institution. And so I'm going to go back to this in a second, but you may want to think directly about making a budget, right? The fact that and let's say there are some goods are more preferred by women. You'd expect descriptive representation to lead to a budget that reflect those preferences along the lines of [inaudible 00:16:18] and the flow for example, right?
But of course, if there are obstacles of this type of the type we described, it may not quite happen, right? So we're hypothesizing all of this here, really. I'm going to show you sort of bits of data to sort of try and test each part of this in a second, right? So to summarize a little bit our hypothesis before I get into the design, we hypothesize that those sarpanches in locations in which discussion rules are explicitly mentioned will have more voice and be more influential. The second point here is that this assignment to treatment will improve their role and enhance their role and increase their influence especially if the sarpanches is from a disadvantaged group. So here are women. So we to put it differently, we expect that there's an interaction effect between gender and the treatment, right?
And the last one is of course, that the assignment to treatment will also have an effect on the second kind of outcome, which is the outcome of the deliberation to sort of a budgetary choice that we ask them to make. So in this sound quite mysterious, how we would test all of this. So I'll tell you briefly about the design. So what we do is we organize a meeting that is similar to the masik sabha. It's not a masik sabha of course, it's not the real masik sabha, but it's inspired. And it mimics the masik sabha, which is the gram panchayat's monthly meeting in which a lot of important decisions are taken, right?
When we organize this meeting, randomly we vary whether rules of discussion are mentioned at the onset of the meeting. And I'll show you that there's another variation in a case. And of course, because we organize the meeting, we have a whole system to code. Well, almost everything that's happening during the meeting, but also observe the decisions that people make as a group, right? So think of this as really as a lab in the field experiment. People have studied gram sabhas a lot. Masik sabhas are a little bit less known. So I just thought I would show you a brief picture of this. It's a small meeting, right? Often it's really a few people in the room. And in that sense, what we manage to organize is really similar in terms of scope and number of people that do participate.
So what we do is... So the study is happening in 600 gram panchayats. I was hoping I could show you full data today for disclosure. We're still super delayed because of COVID. So of course, I'm going to show you data from three out of the five districts in which we're doing this study. There's a lot of implications to that, but big one is that I'm not going to show you significant estimate, because we actually don't have our own data, but I thought it would be just really sad to not show you some data. So bear with me there. Importantly, we sort of in our sampling we block on gender and caste. We really have half of the sample or more that is actually a female sarpanch in our sample, right. Practically, how do we proceed? We first interview three... Well, the village actors individually, right?
Sort of and privately. And once we have done that, we actually ask them to participate to our standardized meeting. And during this meeting, they're asked to decide how to gram panchayat to allocate 10 Lakhs in budget, right? And one important fact I'd like to mention is that because this is a study that is supported by the State Election Commission, we do get overwhelming participation, and we think that people in good faith and they actually really care about the outcomes they work on to sort of obtain to each. So I told you there was a twist in our intervention. It's a bit, and it's as such we vary in fact, not one thing, but really two things. And so we have two treatments not one, right? We vary first, whether rules are mentioned, but also if they are mentioned, we vary whether the sarpanch is asked to speak first? That would be our first treatment T1 or asked to speak last, right?
And remember this is a meeting between three people, and we're asking them to come up with a choice. In the status quo, in the control group, no rules are mentioned. It's that simple. And of course, we sort of recalled everything, I told you some of that. So number of actual... So I'm going to stay in this project. We have a ton of different outcomes, right? I'm going to show you some of them today, feel free to ask us in the Q&A if we're measuring something else that you're thinking about, we probably are. This is a really long instrument. I'll show you the results, this way you'll see exactly what these outcomes we're talking about are. So first I've told you about the absence such significance I'm just going to show you means today, but the first bit of evidence I want to show you is evidence of the gender gap before the experiment, right?
And again, there's a million sort of ways in which we could measure this. I've chosen today a very simple metric, right? Before the experiment, we address this group of three actors and ask them a question about village issues, right? We deliberately ask our interviewers to ask the question to the group and not to a specific individual, right? And one of the things we measure in CIPI whether the sarpanch anticipate to this question asked to the group, right? And as you see here in about 70 something percent of the villages, the sarpanch speaks, right? Now, that's of course a pooled result. We have both men and women in there. I'd like to show you now, and this is where the gender gap is the difference between men and women, right? The point here is simple. When you ask a question and many of you who've done field work in Rural India will be familiar with that.
Men are almost always participate to the discussion. Women, it's much more of a mixed back, right? So there a huge gap here in there. This is of course, one piece of data that I must interest of time, I'm going to show you other pre-experimental data here, but know that we have other estimates of this type about the sar so speaking time, about whether how people are positioned in the room, et cetera, that sort of kind of collaborate this as well as a more subjective estimates that we collect. And in other parts of the survey. So I'm going to sort of zoom in on that, but basically I'm going to sort of go a little bit faster, but the point here is of course, as a gender gap, right? We sort of kind of intuitively knew that, we've now measured it. And so what I want to show you now is what the effect of the interventions are, right?
And remember there are two interventions. Treatment one here is the treatment in which rules are mentioned, but sarpanch speak first or is asked to speak first. And treatment two, the sarpanch is to speak last, right? And here in the control, no rules are mentioned, right? So here again, this is the pooled results. I'm first showing you whether people complied with our request, right? And it should be clear. They did not always comply with our request. So you may want to think of this as a purely as an intention to treat kind of effect in experimental lingo, but there is still a difference between the control and the treatment here, right? Interestingly, the treatment two and the sort of control are very similar, but you see that the results on other outcomes are actually not, right?
So point here is probabilistically are intervention is followed, right? Is implemented, but not always. Excuse me. Now again, I want to show you a piece of data about this sort of gender gap, we have men on the left and women on the right. Men here we're simply looking at treatment one, right? In which we ask the sarpanch to speak first upon sort of nicely making the request on behalf of an important team of researcher that are working for the State Election Commission. We do find that women are more likely than aware probably to sort of participate, but still it's not a very strong compliance, right?
Again, it's an increase on the baseline, but it's not a very strong compliance when you think about the [inaudible 00:26:02]. Okay. Who speaks during the discussion about the budget? We get to the discussion about the budget, that's what follows. We ask them to make a choice between five policy priorities. How would you divide 10 lakhs in between five things? And so here in this graph, we simply measure the percentage of the time we have the data that each of the actors spoke and we break it down by gender, right? So we have here treatment one, treatment two, et cetera. So as you may see, it appears to be increasing both among men and women. Of course, for men, the baseline level is much higher, right? The percentage of time that they speak, right?
So it's not only that they speak first or that they speak last, it's also that they speak a bit more, right? There's something consistent here basically across the study, which is treatment two in which women, or in which the sarpanch that speak last appears to have a much larger effect than treatment one, right? We're pretty agnostic about that in our sort of theoretical section, but you'll see that it's coming. It's not the same thing for men, we don't really know why it's so far so much of very much a question mark.
Next piece of data, we ask our interviewers to think about subjectively, but we give them some pretty precise guidelines about how to think about that. Who they perceived to have made the final decision in the budget allocation decision, right? And here in blue, you have the percentage of time that they answer the sarpanch, right? So first off, first thing I want to mention here is that in the baseline, this rate is extremely low, right? You would expect the village council president to be the person that makes the decision much more often than that, right? Remember there are three actors. So if they were equally likely to be the final, the key decision maker, you would expect this to be at least 10 points higher, right? Here is very low, and that's going to sort of go back to another sort of results we get elsewhere, which is that sarpanches overall are not necessarily as influential as it should be.
But nonetheless, the other sort of big thing that should jump at you here is that just mentioning these discussion rules, whether they actually mention that the sarpanch should speak first or last appears to really increase the perception that the sarpanch is making the final decision. So I'm sort of predicting opposition to the subjective measures. So I want to show you the next piece of data. Oh, here we have... Well, I'll keep on this for now, but maybe it'll be useful in the training. I want to show you the next piece of data, which is this. And I think this is probably the important one in which we measure whether the post treatment budget priority. So the thing that the group decides should be allocated the largest amount of money matches what the sarpanch privately thinks should be allocated the most amount of money.
Something I think I forgot to tell you is that in the pre experiment, during those private interviews, we also measure the private preferences of these actors, right? This way we're able to compare the private preferences, pre experiment to the outcome of the group discussion, right? So we have a number of self measures to a different measures here using the simplest of all of our measures to see whether the budget allocation resembles the budget allocation of the sarpanch, which is whether the number one priority is also the number one priority. Of course, all other measures I'm happy to discuss in the training. But by and large, you see also here that there's a big increase in the group deciding to invest or the most in a topic or in a theme that E that matches what the sarpanch wanted, right?
So you also have a pretty big effect. These are the pool effect. I'm going to show you something which is much more interesting, which is the effect by gender, which is probably our most important result. If you remembered one graph, which would've to be this one, I think, right? So here you see two things, right? You see of course that the effect of the treatment one and treatment two appears to have an effect. Same as above treatment one is more powerful for men, but treatment two is more powerful for women, right? We may want to think about why that is, but in both cases, we do find that mentioning discussion rules actually provides more influence to the sarpanch on the final choice made the group.
Something else we can do with that cool data is actually look at what happens right after the experiment, right? When discussion rules don't really apply anymore, because the budget decisions we made, but we ask them a series of other question, and we can simply measure us overall speaking time, right? The percentage of time that the sarpanch speaks throughout the meeting, right? So a very small portion of this is actually the experiment, because this is the whole interview. The whole interview is much longer than the experiment itself.
One thing that is encouraging is this. So in red you have the mean level in the control group. So at the bottom for women at the top for men, right? So this seems to be indicating a little bit of a reduction in the gender gap and there's intervention between across men and women, right? So the treatment here reduce the percentage of time that men speak and they appear to increase it a little bit for women, which is actually what we would expect. Not that this is not necessarily what we find on the other outcomes I just showed you like this one here, you have an about has big an effect on men as on women, which again is something to think about in interpreting these results. One last piece of data I want to show you before I sort of jump to conclusion is this.
And because this is also a sobering kind of results. We also, we're curious to see whether after implementing these discussion rules, there would be something of a backlash, right? And so once you've had to sort of let the person speak, are you going to be bad to them later when the rules don't apply anymore? Right? And I mean, sadly, we appear to be finding something along these lines, right? And what this show is simply that both for men and especially for women if you compare to the control, not big effects, but interruptions appear to be much more frequent, I mean, somewhat more frequent, I shouldn't say much more frequent. Let's see again, how much that falls once we have our entire sample. So I have two slides left. I'm just going to give you a few tentative sort of conclusions just to summarize all of this, because it was a lot of different slides and tell you what we think these shows and what we think we can say about this and not, right?
So the first thing is that the intervention and I think appears to have an effect on how central the role sarpanch play in decision making process, right? This is not necessarily exceptional as a finding in so far that the rules are meant to do that, right? And so we do get some compliance. The interesting thing though, right? So sarpanch speak more often and those treatments and they're more key to decision making, et cetera. Now, what I want to highlight here is that this is still something not to be taken for granted given how many of the people actually did not follow the request, right? So here I haven't showed you results broken down by whether the treatment was actually implemented, and whether people listened to the instructions, right?
We still do find that a mere request to sort of institute these discussion rules did that. I think there's something interesting there already. Second, it appears to have an on the outcome of deliberation. I told you that probably is a more interesting and a more important results. The outcomes of the deliberation now better reflects the preferences of the sarpanch, right? And remember typically in the setup of the study, we've measured the preferences of the sarpanch privately a day before when these actors were not aware that there would be a meeting of this type, right? And again, they don't know that there's going to be a comparison between the outcome and their private preferences that they expressed. So there's something really nice about this. Three, this is more of a question mark, and I should say T2 consistently more effective, consequential than T1 for women, right?
So speaking last appears to have a bigger effect on women speaking first maybe on men. Again, we don't have a full sort of theory as to why one of these interventions should be more efficient than the other, but we do find this difference between men and women interesting. Hypothesis they are sort of welcome. And however, as so again, what I mentioned, I think we have sort of potentially problematic or unintended consequences in there, right? So problematic, let's see if that sticks, but there seems to be that interruptions are up afterwards, so that would be a problem. And then intended is something to sort of think about a little more, but the intervention does not necessarily close the gender gap in influence across villages if the key measure is the budget decision, right?
And the reason for that is simply that our initial assumption that male sarpanch don't need the intervention to speak more and to influence is wrong, to put it simply, right? So in fact, they also react strongly or even as strong are maybe more strong than women speaking to the intervention and the benefit from it in the budget discussion, right? And to us, these points to the fact that... So not only our initial assumption that they don't need that is wrong. And remember among the group of men, you're going to have for example, valid men who are also victim to this kind of interruption and sort of proxies that we observe in the case of women, right? So it's not necessarily complete surprise, but maybe at this point as a problem more generally between the bureaucratic actors and the elected actors, rather than between female elected actors and other actors.
So last word, so to think a little bit about policy based on this is a little bit difficult of course, because we don't necessarily have all the data we would want to have to be able to say something definite chief about this, but the current data I've shown you, I suggest at least that the change in procedural rules if they were enforced, could be beneficial to both male and female sarpanch and help them regain voice and centrality and influence in decision making processes, right? However, this is not going to reduce the gender gap. Here an intervention actually gives E, oh, so not doing to reduce the gender gap across villages, right? One thing that we haven't sort of completely prepared for is that the intervention is supposed to be a show to both men and female versus bureaucratic actors, but not necessarily in a way that is actually closing the gender gap between that. I'll stop right here. Thanks. Should I stop sharing?
Naveen Bharathi:
Yeah, you can Simon.
Simon Chauchard:
Okay.
Naveen Bharathi:
Yeah. Thanks a lot for this presentation. And it's very interesting. I'll have a couple of clarification questions on the research design. Do the members work for any decision? Always the panchayat pradhan or sarpanch take the decision? Or if it's a voting process, how do you really measure that the sarpanch is taking the final list? The second question I had was-
Simon Chauchard:
Should I answer? Okay, go ahead.
Naveen Bharathi:
I'll finish this then. The second question I had was this was not the panchayat meeting. So when you mean the budget discussion, did you give them a particular part of the budgetary location to discuss, or was it like they could select the topic or was it like... How did that happen? These two are my questions.
Simon Chauchard:
Okay. This was-
Naveen Bharathi:
Yeah, an experimental setting. This is not the real gram panchayat meeting, the monthly meeting, right?
Simon Chauchard:
No. So-
Naveen Bharathi:
These are all my questions. These two are my-
Simon Chauchard:
So yes. So it's a good reminders. So what we're trying to do is it is very much an experiment and there's a degree of our specialty that is big in fact, right? So what we're trying to do is we're trying to mimic something that resembles the masik sabha, right? So I think one thing we're happy about in the way in which we produce the way a decision is taken is that there isn't a formal vote in our process either, right? And that's part of the problem. We identify the onset. Often there's a number of people around the table. We need to decide whether we go for A or B people speak, and then the decision is taken. And it's not very clear how, right? And that's exactly what we aim to do here.
And the variation we introduce is a variation in whether we actually introduce formal discussion groups or not. But the budget question is to give you more detail on this, we ask them so we give them, which we tel them there are 10 lacks to distribute, right? If you had 10 lacks to distribute, which policy priorities would you choose, and how much would you allocate to each other? Again, it's a hypothetical decision. And again, this is not the real meeting, but we think that it's the decision that matters again, because this comes from the State Election Commission. And there's some degree of seriousness in the process. And people do think about what they're reporting, right? But again, it's an artificial sort of experimental setup. Totally.
Naveen Bharathi:
Got it. Yeah. Tariq, you had some questions.
Tariq Thachil:
Well, I had a question, but I'm happy to ask it while other people are asking this. First of all, thanks so much for presenting that. And it's great to have I think Rachel and Alyssa both on the call. So all of you guys, it's fun to see this project and how you guys are persevered through COVID to even collect three fifths of it, is really impressive. And I'm looking forward to seeing the full results. I guess I have a broad question of kind of not being like very well steeped in this particular show, but like the framing I understood to really focus on deliberation. And I guess, in my understanding, what I'm seeing is really a focus on centralized decision making, which isn't really the same thing as deliberation at least to me.
So one, why should we want? I understand from a decision making point of view that maybe often what happens, and I understand that if women are sarpanches that what some of the fallout of that could be, but I'm not sure just normatively speaking, why do we want to centralize decision making? And how is that in a framework of democratic deliberation?
The kind of the kind of outcome of interest. And I guess related to that is I was thinking of figure nine in particular, there were a couple of figures. I mean, obviously there's a gender gap like figure nine doesn't want you to know you pointed out was especially important. In that figure, if I see... I was quite struck that there's actually quite a lot of at baseline, there's not I think like a five percentage point difference was 33 to 28, but so that seemed quite, I mean, I don't know what your priors were, but it did seem like women sarpanches did often have, or at least the gap was not as a significant as we wanted. And so in that the treatment seems to be centralizing the authority figure. And maybe that baseline was actually more, I don't know, deliberative if the decision wasn't clear who took that.
And so I guess I could even understand if you did the intervention in a way that you were kind of later on really focusing on women sarpanches, but if we institutionalize the procedure where we are strengthening the office of the sarpanch, and then that office is often occupied by men. And you guys are obviously thinking about all of this at the end, but I just wanted you to kind of think a little bit, but for me, that was really kind of the big takeaway in terms of how I'm thinking about the kind of theoretical motivation behind this and some of the normative implications. So just any of your thoughts on that would be great.
Simon Chauchard:
Yeah. That's great. Thanks, Tariq. So I mean, the goal is not actually to centralize, it is the sort of metric where after that motivates us is the gender gap, right? And the fact that men are much more likely that women to participate in decision making seems dramatically problematic to us, right? What I mentioned at the end is by the way also problematic, right? Even if you think that deliberation is important, and I do agree with you on this, there's a democratic problem if we do systematically find that one of the most decisive actor is a bureaucrat, right? So in that sense, there's a problem as well, which is not at all sort of granted it's not what we were sort of going for, but this is what we seem to find, right?
But, I mean, I think what should be the ideal level of participation on all of these metrics? It's hard to tell, right? True. What we are looking for is probably for an integration that narrows the gender gap, whatever the baseline level is, it's not what we find here. So again, I think thinking through that, I agree sort of super important. And another thing sort of deliberation is a really big word, right? And so anyone who's sort of attended gram panchayat meeting has seen very little deliberation. There's some discussion. There's something that looks like a deliberation in some cases, but again, it's sort of very diverse, but it's also not true to say that it's always centralized, it's just really diverse is what I think we've learned on our own there. But, yeah. I agree on the label it's not ideal. Thanks.
Naveen Bharathi:
You want to go next?
Tariq Thachil:
I don't think the name was clear Naveen. Can you just repeat who you asking?
Naveen Bharathi:
Yeah. He's here. Gilles, can you ask your question next?
Gilles Verniers:
Yeah. Hi everyone. Hi.
Simon Chauchard:
Hello.
Gilles Verniers:
Thank you so much. And yeah, thank you very much for this presentation. It's a broad question. If I understand you present yourself in villages close with the dual legitimacy and authority of researchers, but as well with the endorsement of the State Election Commission. And so it might be fair to assume that the experiment is conducted under the [inaudible 00:47:28] authority. And so how is it likely that the effect that you induce through your intervention will sustain itself? Should women pradhan attempt to enforce the deliberation rule after you've gone? My question, and I'm sure that you have thought about it and I'm really curious to see what your answer might be is, I'm wondering to what extent the effect is not the rule, but the inducement itself and your presence itself and the condition in which you present yourself as a means to get the rule implemented for a time where you are present? Thank you.
Simon Chauchard:
Thanks Gilles. Yeah. Those are great questions where we've been of course, thinking about. So the duration is of course, a big concern, right? I think in this setup, we don't have much else than... I mean, the one thing we're able to show is that it lasts a little bit, whether it lasts a little bit longer during the-
Gilles Verniers:
Not after you're gone. While you're there.
Simon Chauchard:
No, no, exactly not after we're gone. No, we agree. That's what I was saying. We do this literally in the same afternoon, the morning, right? So we don't have that. However, I think what we're going, what we're trying to show is that is whether actually these discussion rules had an effect in the short term is actually interesting to know in and of itself, I think, right? It wasn't definitely, it wasn't clear to us at the onset that it would actually have an effect, right? And then I think you may want to think about the difference between the different outcomes, right?
It'd be very easy for the group to follow the rules and let the person speak, and then reach a decision at the end of a discussion that does not reflect necessarily their preferences, right? That seems much more unclear to me that the intervention would've a direct effect on that for the reason you mentioned, but again, I think it's a tricky question. On the mechanisms, yeah. So I think there's at least two things that are possible. One is indeed there is the nice story I've painted on my slides, which is persuasion happens and so on, because the sarpanch speaks more, but the alternative of course, is something else which is perceptions of social norms change or something like this or people are trying to behave because there's an outsider in there. This is something we're going to sort of work on. We have some survey items we haven't looked at yet, but me allows to sort of get that, but again, that's a hard question. So all very good points.
Naveen Bharathi:
We have a question by RO. Do you want to go next?
Rochana Bajpai:
Hi. Thanks, Naveen. Sorry. Really interesting work. And I don't come to this as a political scientist-
Simon Chauchard:
I thought so.
Rochana Bajpai:
... but... Hi. Sorry, I can't put my video on at the moment.
Simon Chauchard:
No worries.
Rochana Bajpai:
Hugely interesting and great to see this kind of work being done. I come to it as a political theorist, not as a political scientist. So I'm not sort of familiar with the methodological aspects, but I had similar questions to ones which Tariq was raising and specifically in relation to deliberation. First, how attached are you to that term? Because in our broader discipline, certainly of politic science, it has this specific meaning. And then it didn't seem to me that you were trying to track the effects of deliberation. I mean on say preferences.
Simon Chauchard:
No.
Rochana Bajpai:
You would assume if that was what you were doing, that you would look at preferences before and then see what, how her own and how the sarpanches own preferences changed through that.
I mean, maybe I missed it, but I don't think that's what you interested in doing, right? So that's one of the things.
Simon Chauchard:
No.
Rochana Bajpai:
And in case, so why burden yourself with a lot of assumptions which come with deliberation, including its effect on persuasion, including that this is a deliberative forum in the first place and so on. So those were sort of one set of questions. The second is and I come completely share your sort of aim, but how are you measuring the... Is it... And what are you terming this sort of making the objective that is to make women's? is it to see how substantively how much descriptive representation leads to not substantive representation in terms of policies that are pro women, but more in terms of how much role they have? So what is it that you're trying to track the dependent? Because they seem to be two, three different ways of doing that as well in terms of what, how you're measuring, if you like that influence that sarpanches have essentially. So I'll stop there.
Simon Chauchard:
Thanks the questions. Those are great too. So on deliberation, I think what I would say here is get a pushback on this term. And I sort of concur with quite a lot of that. Her intervention does not actually imply or encourage any kind of deliberation, right? It encourages discussion rules that are inclusive in some way, right? If I speak be first and you speak second and someone speaks third, and at the end of that we say, "Here is our choice." This is not deliberation. Completely agree, right? So there I think you could think by the way about a slightly different intervention, but it gets at that, right? This is not exactly what you're doing. So I take your point. On the dependent variable the metric we're interested in, so we're interested in all of this, basically we're interested in the whole causal chain, right?
We're interested in whether people speak, whether they are present first during meetings and sort of, we also measure that in a much more observational manner, right? Also whether they participate to the debate, how much they participate to the debate, but also in the outcome of the discussion if we don't want to use the word deliberation, right? And the reason why I think we should be interested in both of these things is while the outcome reads it's kind of evident I think why we should be interested in that, which is we have quotas so that women are elected on the basis of a big assumption that women may have different preferences. So we want the preferences to be different if women are in office, right? Now automatically I think there's an easy case for that.
Now. Why obsessed about voice and speaking time and all of these other metrics? Well, because these are interesting in their own way. And they're just... I think empirically you could if you know what I mean, first of all, it's important that everybody elected is able to speak. And again, the question of the level, the different types question is this sort of in the air. But you may also want to think about the consequences of how having a system in which a 100 K women are elected and a good share of them never speaks, right? The repercussions on this on other women are just probably extremely important, I would argue. And I find that this of this course about this sarpanch party, et cetera, very problematic for the sort of psychological effect that these quotes as are sort of supposed to have as well, right? So we're interested in both is what I would say.
Rochana Bajpai:
Thank you. But do you see the latter as substantive representation or what are you defining your dependent variable as?
Simon Chauchard:
Yes. Naveen, is that?
Naveen Bharathi:
Yeah. You can quickly answer that. We have a lot of time.
Simon Chauchard:
Okay. No it's not necessarily substantive representation, but well, it is related to the idea of substantive representation, right? Again, it does an elected official manage to influence the outcome of a collective process, right? So description of the idea of this of substantive representation is related to that I would say, right? But, yeah.
Naveen Bharathi:
Thanks. We have Rithika and Surili. Can you ask these questions quickly both of you?
Rithika Kumar:
Yeah. I'm going to be really quick. Thanks Simon and nice to see Rachel and Alyssa as well. My question was basically, I just wanted to know firstly about like the nature of this meeting and as a female sarpanch, does she come by herself? Is her husband around and things like that? The other thing is what was the idea behind splitting your treatments into speaking first, who was the speaking last. What was the rationale behind actually going in with this kind of treatment? And also what are the ways in which you are interpreting this data? Because a lot of the data collection is kind of based on how subjectively the enumerators evaluate the situation. So if you can talk a bit about this as well. Thanks.
Naveen Bharathi:
Surili, do you want add to that?
Simon Chauchard:
That's really good questions.
Surili Sheth:
[inaudible 00:58:07], should I go ahead and ask?
Naveen Bharathi:
Yeah, please.
Surili Sheth:
[crosstalk 00:58:10].
Naveen Bharathi:
Okay. Yeah.
Surili Sheth:
Actually mine is a pretty redundant, so I'll just echo I think what Gille's question was, which was I wanted to ask more about the research design and specifically what the treatment actually is. I was also thinking whether the treatment is about a change in deliberation rules or about who asked. So a research team with the election commission, everyone in the room knows this. And if it's the latter, you could think, oh, maybe you could have that same setup of who asks, asking something way more direct to structure it even more towards reducing a gender gap. And so I... But I think that was asked, but interested to hear if you have anything more to say on that.
Naveen Bharathi:
Yeah. One more question just to add to this by [Somnia 00:59:00]. So was this discussion made compulsory by the election commission or was it like voluntary by the gram panchayats to participate?
Simon Chauchard:
Voluntary.
Naveen Bharathi:
Can you answer Rithika's questions?
Simon Chauchard:
Okay. So I didn't actually, but it's okay. So there are lots of question Rithika, I'm not going to go through all of it, but on sort of how we code this. Yes, we do rely on our team to recode these things, right? So we have empirically one of the things we do is so we have a team of enumerator when the meeting happens. One of them asks the question and leads the interview, and the other one is really coder present there, right? So now some of these things are just very subjective, right? So who do you think did that? Who's the most made the final decision is completely subjective kind of coding? Who spoke first is not, right? Percentage of time that they spoke is not. We also have recordings of these things oh, and one more important thing this is by the way a gender balanced team, right?
There's another potential variation we're going to be looking at about the gender of the person leading the meeting, which is not, I haven't talked about yet, but is another part of this, right? So one more precision, I forget who asked that, but so the election commission is not in the room, right? This is to be clear, right? That would be a bit intimidating. And there indeed really wouldn't trust any of this, right? No, we simply have a letter of support from the State Election Commission saying, "These are what people consider participating." And so when we ask people to participate, we really get a very, very high participation rate after quite a bit of insistence that people take part in the collective interview, but we don't necessarily feel that people are sort of over impressed or sort of feel that they have gun over their head during those meetings. Actually some of the numbers in there, I think show that social desirability in sort of really, really limited in some of the patterns we actually describe.
Rithika, I feel I'm forgetting stuff, but I should keep asking questions. So I'll get back to you separately if that's okay.
Rithika Kumar:
Sounds good.
Naveen Bharathi:
Here we have the last set of questions by Tanushree. Tanushree, do you want to go next? And after that, we have a lot of questions and I'll ask them to write directly to you. So Tanushree, do you want to go next?
Simon Chauchard:
Sounds good.
Naveen Bharathi:
Please be brief because we have already run out of time. So keep it as brief as possible.
Tanushree Goyal:
A lot of pressure, but thanks Simon for this really interesting presentation. This is definitely a great project. I don't want to get into the large sort of questions, Tariq and Rochana raised. I just want to echo them. And I think apart from that, I mean, just focusing on your experiments, even if you part these issues aside, I wonder the more I see your responses on the chat, the more I like there's so much going on in the experiment that you would like to know more about, because my worry is that sort of everything drives me to think that you are overestimating your effect. Lots of things like so the surveyors are doing some objective measure. They know where the rule is present and where the rule is absent. So what have you done to overcome that sort of like subconscious bias that can also exist?
So for instance, where women are recruited without, or both are recruited without the institutional rules, are women more discouraged from speaking? Because we know that when women feel they're observed, they're less likely to speak. So there's this discouragement effect, which will also pick up from your treatment. So there's so much going on. And I think my suggestion for future rations would be to include some manipulation checks to so that we are less worried about these sort of questions or to have a placebo treatment where you do not worry, maybe where the institutional rules are absent, but you know what else is going on along with some other treatment, which you don't expect to move anything. So I think that those should be the things that you should also show to at least just convince us that even this sort of like artificial setup, lab is actually moving something at all. So that could be a case for future these kind of big questions, but I think you need to first establish to us that this is actually something that is moving by the treatment. Thank you.
Naveen Bharathi:
So one more question by [inaudible 01:04:01]. I'm going to just add to this. So he's asking that, how did you train the interviewers? How did you train the people who moderated these sessions?
Simon Chauchard:
Woof. That's a long topic. We train them at length. There are sort of code book for this project is really long. So I don't know how much time I have to sort of get into that. There's tons of rules about what to say what not to say and so on. So I just want to sort of speak about sort of go back to Tanushree's actually, because I'm not sure what the manipulation check would look like actually, other than what I've showed you, which is that whether people actually follow the recommendation to speak first or last or not, right? And so what seems significant to me is that we do get those results.
And again, significance is up in there at the moment, but we do get a difference in the context in which broadly half of the people don't even follow the recommendation, right? And the other thing is, if you think about the control group, the woman present in the room in the control group is also under observation from team of surveyors, right? It really doesn't change anything on that front, right? So I'm not sure I completely see the difference there between the treatment and the control, right? But maybe I misunderstood the comment.
Naveen Bharathi:
Yeah. We have run out of time. We have a lot of questions. So I ask everybody to write directly to Simon and I'll also send you the chat box to you directly.
Simon Chauchard:
Ah, yes. Can I answer Rithika's question that I now see in the chat? Sorry.
Naveen Bharathi:
Huh?
Tariq Thachil:
Yeah, you can. Why don't you go for it? I mean, if people need to leave, they can, but why don't you go ahead?
Simon Chauchard:
Yeah. So the difference by the two treatments. What I mean pretty agnostic about that actually. There's a lot of sort of ex substantial rising came into this, to be honest with you. That just so of really good arguments for why one would be more effective than the other on both sides I think. Speaking last, I mean, you can think about it's easier to sort of have the last word and sort of be more influential if you're the last one speak. People sort of study parliamentary politics and all that. And there's lots of studies about the order in which people speak in assemblies. Speaking first is akin to something like agenda setting maybe, right? Or sort of enunciating sort of the rational that should guide the choice or something like that, right? So think those are good arguments, both sides. And we're pretty exhaust and sort of figure that the end was hopefully large enough for us to do that, but yeah, it's fishing I would say.
Naveen Bharathi:
Yeah. I will send the chat box to you Simon so we can-
Simon Chauchard:
Yeah. I see. That would be lovely.
Naveen Bharathi:
There are lots of questions and mostly to do with the design and how are you going to measure all that? So thanks a lot everyone for attending this seminar. Thanks Simon. Thanks Rachel. Thanks everyone. And so next week we don't have the seminar. We're having the spring break. Our seminar is from 14 days from today. So please register in our website for the seminar, and thanks a lot and have a good day.
Tariq Thachil:
Thanks so much for coming all three of you coming on Zoom. It was great to hear.
Simon Chauchard:
Thanks.
Naveen Bharathi:
I'm sorry, I missed Alyssa's name. Sorry for that.
Tariq Thachil:
No, thanks Alyssa. Thanks Rachel for joining. Thanks Simon.