Culture & Society

Ideas of India: What Data Can and Cannot Tell Us

Shruti Rajagopalan and Rukmini S. discuss the difficulties of collecting and interpreting data about Indian populations

Data on the coronavirus and other issues don’t always tell the whole story. Image Credit: bgblue/Getty Images

Ideas of India is a podcast in which Mercatus Senior Research Fellow Shruti Rajagopalan examines the academic ideas that can propel India forward. You can subscribe to the podcast on AppleSpotifyGoogleOvercastStitcher or the podcast app of your choice.

In this episode, Shruti speaks with Rukmini about her book, “Whole Numbers and Half Truths: What Data Can and Cannot Tell Us About Modern India.” They discuss Indians’ multifaceted identities, endogamy, preference falsification and much more. Rukmini is a data journalist who writes columns for Mint, IndiaSpend and other publications. She has also written for HuffPost India, The Hindu and The Times of India. Additionally, she hosts a pandemic mini-podcast called The Moving Curve.

SHRUTI RAJAGOPALAN: Welcome to Ideas of India, a podcast where we examine academic ideas that can propel India forward. My name is Shruti Rajagopalan. Today my guest is Rukmini Shrinivasan, an independent data journalist based in Chennai and the author of the book “Whole Numbers and Half Truths: What Data Can and Cannot Tell Us About Modern India.”

We spoke about how well the data captures the complex and fraught lives of Indians, segregation and endogamy, the politics of political data, excess mortality during COVID-19, preference falsification and 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 Discourse Magazine DOT COM.

Hi, Rukmini. Thanks so much for doing this. Welcome to the show.

RUKMINI SHRINIVASAN: Thanks for having me, Shruti.

Segregation in Life and Numbers

SHRUTI: One thing that really jumped out when I was reading the book was just how segregated Indians are. We throw around the word Indians. Of course, there’s a geographical boundary, and then we talk about South Indian states versus North Indian states or Hindi-speaking versus another language. First of all, each of these categories is very large. A Hindi-speaking belt—we’re talking about the size of Brazil plus-plus-plus, right?

RUKMINI: We’re talking about dozens and dozens of languages, each of them spoken by several million people, not just one language.

SHRUTI: Exactly. One is, the segregation that we have is so deep. My lived experience in India is, you can go your entire life without meeting people who are not like oneself. Most of my childhood growing up in India, I didn’t come across too many people who were a different caste or a different religion. If I did, it was in passing; these were not intimate relationships, and they weren’t deeply entangled. That’s not unique to me; this is true across caste and socioeconomic class and privilege. There is a tendency to hang with people like oneself in India.

Are we getting all the categorizations completely wrong, in the way we think about collecting data in India, when actually we are segregating along completely different lines, which have relatively less to do with geographical boundary or district boundary or districts where you have people above poverty line and below or with big Adivasi populations or not, and so on?

RUKMINI: Yes, you do see this quite starkly played out. Just this morning, The Hindu has a photo of a street in a district that borders Pondicherry. There’s a Sunday lockdown in Tamil Nadu, while there are no Sunday lockdowns in Pondicherry, for example. One side of the street has shops shuttered and the other side doesn’t. You come across countless examples of the way boundaries intersect with people’s lives. You come across it in very physical manifestations all the time.

I do agree that we lead deeply, deeply segregated lives. One of the things—I don’t know if this is common to people, to other journalists as well. But across the country, and even now actually here in Chennai, because I’m not a native Tamil speaker—so my Tamil obviously sounds wrong, I’m conjugating verbs wrongly all the time—I get asked if I’m a foreigner very often.

I used to think, when I used to travel in the north to report, I used to think that it was the Hindi word “pardesi,” and that was the equivalent, and they didn’t actually mean foreigner. But people in multiple parts of the country, while I’m reporting, have asked me if I’m from another country and not from another part of the country. I was once in Jharkhand, and at that time, I was doing some election reporting, and I was in a village of people where all of the young men were out-migrants, and many of them, many of the sons of the village were in Chennai. They were construction laborers in Chennai.

I hadn’t yet moved to Chennai then, but I was already married, and my husband is from Chennai. So I gave them that, because I think I’d recently traveled to Chennai, and I said, “You know what? I went there recently.” The questions they had about Chennai were things like, what sort of gods do they worship there, and what is worship like? It truly was people trying to imagine a completely foreign land. Yes, it’s amazing.

When you think even about the statistics of how rare it is to live outside the village or district of your birth, making these casual, big journeys across the country . . . with so many of us, with the privilege of mobility, we don’t realize that—just how rare it is to move outside the area that you’ve grown up in. Whether these are meaningful categories, I suppose they do show how rare boundary-crossing across these categories is.

Of course, the most common example of lack of boundary-crossing is around intercaste marriage. It’s been surprising to me how surprising it is to other people that intercaste marriage is so uncommon. This is one of those things, because I’ve used the same dataset for years and years now, I didn’t know that people didn’t realize that it was such a small number.

Yes, I do think that it’s something we should be talking more about, not just the fact that there is such little intercaste marriage, but also the fact that there is such little support for the very notion of intercaste marriage. Let alone having one yourself, the fact that no one should have an intercaste marriage, and interreligious marriages in particular, really says something about boundary-crossing.

Again, here, there’s a dangerous narrative that’s crept in in the last couple of years among liberals, which is that people may not do this sort of boundary-crossing in their own personal lives, but they are live-and-let-live on a broader scale. And they’re okay with the notional concept of a Muslim family, they just don’t necessarily want someone next door.

I don’t think that that’s a useful or accurate way to be reading this, and this lack of boundary-crossing says a lot more. It says these segregated lives don’t go alongside notional ideas of equality. I don’t think they accompany those at all. Yes, I do think that the categorizations that have become default in surveys have ossified over time and do need to change. I see this particularly when it comes to election surveys. The same four basic religious groups and five basic caste groups, and that’s it, and then you’re done, is really problematic.

Again, as we know from urbanization data, talking about rural versus urban is so problematic in India with a huge spectrum in between. Even talking about things like public versus private is pointless, again, when you talk about healthcare or even credential providers versus non-credential providers. Yes, we’re pretty bad at capturing spectrums, and some of our categories, just the very fact that the census categories of employment—it’s agricultural, home-based, artisanal and other, or something like that. There is such little—it’s shocking that we know so little about non-agricultural work, through the census at least.

SHRUTI: Or even agricultural work, because most of it is in and around the farm. But the number of people who are actually, directly in agricultural work as farmers is a fraction of people who say that they’re in agricultural work.

RUKMINI: Right. Also, what we don’t capture well at all is mobility between categories for all sorts of things. We don’t capture physical mobility. Chinmay [Tumbe] talks about seasonal migration and how poorly we capture that short-term seasonal migration. Similarly, moving in and out of jobs and job categories is something that we capture very poorly. Yes, we do have these ossified categorizations.

Endogamy

SHRUTI: You know, two things occurred to me when you were speaking. I had written a column once about how we have “Amar Akbar Anthony” secularism in India. Our idea of secularism is that there has to be some kind of brotherhood between people of different religions, and it’s a very post-partition kind of secularism, that you don’t fight among yourselves and there’s no violence and things like that. But each one of them has the good sense to marry, fall in love with and marry within their own religion.

The real Amar Akbar Anthony would have been if Amar is getting married to Jenny or something like that, but that’s not what’s going on in the movie. That is about as far as we are comfortable pushing. Anytime you cross that boundary, then you know exactly what happens. You suddenly see what we call the illiberal faction of India, but actually, it’s most Indians who are deeply uncomfortable with showing interfaith or intercaste marriage or boundary-crossing in any way. You would think this trend is disappearing among the young, but actually, it’s not really.

The only thing that’s changed a little bit is, now they say that you can fall in love with someone from across the boundaries, so to speak, but we know that you can’t marry them. There are very few people for whom crossing that line actually works out as a happy ending, whether it comes to family or the community at large or run-ins with the police and things like that.

RUKMINI: Right. Yes, the change that we do see is in our preferences, but we are not seeing it in actual numbers. I have huge sympathy and empathy for people attempting these relationships because it’s physical danger. This is very fraught territory, and you don’t need to go very far, you don’t need to read books or talk to people very far out of your own circle of friends even to know how hard this boundary-crossing is. I don’t think that there’s a braver act that you can do in India today than pursue a relationship or try to marry someone of another faith or caste. You see it all around you that there are very physical, very real and terrifying repercussions.

This is something that crops up not only in this data, it also crops up in the police data that I looked at where—and this is something now many lawyers tell me that they were noticing it in courts over time. But I saw this in the data, which is that police-supported parental criminalization of intercaste and interreligious relationships is very much a part of our police statistics and our court system.

Given the state backing of coercive laws, or guidelines at least, against discouraging interfaith relationships, we are only going to see more of this. In fact, we are already beginning to see more of this in some states. Yes, I’d say that anyone who attempts to pursue an interfaith or intercaste relationship in India today is committing an act of great bravery.

SHRUTI: This also has unintended consequences, right? Like there’s this big discussion right now over love jihad and things like that. A lot of the women actually convert to the other religion of their partner just to save the hassle of not having to go through the Special Marriage Act, because you know that you need to give 30 days’ notice, which means the police will come calling and things like that. It’s almost perverse, right?

The people who are really policing these interfaith couples are now actually driving them underground, which leads to an unintended consequence of religious conversion, which they are even more against than they are with interfaith marriage. Now it’s become very political because they tend to think that men from a religion are trying to woo women from a different religion and forcing them to convert.

Whereas the reality of Indian life is it’s usually the woman who switches her last name, very often her first name, where she lives. If a regular person just looked at the number of religious conversions within a romantic or intimate relationship, you’d think this is a huge thing going on, until you zoom out and you look at how hard it is to actually get married under the Special Marriage Act for an interfaith couple.

RUKMINI: Right. The fact that that law actually had a provision—and I say “had” because the Allahabad high court modified it, but I can’t say that I know that it has the force of law for the rest of the country—but that it had a provision that the names and the photos had to be displayed on the notice board. I know of friends myself who have been harassed with this.

It’s shocking that it was particularly hard to use it. Yes, there’s a couple I’ve spoken to, for the book as well, who were pursued across the country by their families because it was an interfaith relationship. When they finally got married in Mumbai, they decided not to register their marriage at all. What are we doing to data and registrations of marriages? What is the incentive to do that when you know how dangerous that can be?

Endogamy and Structural Transformation

SHRUTI: You don’t see richer people necessarily having more interfaith relationships or having more liberal values or being less casteist and so on. The standard theory of structural transformation and long-run economic growth is, as incomes increase you actually tend to go in those directions.

Now, is it that we haven’t been in this process of structural transformation long enough that you see that these institutions in India are not breaking down? I mean, endogamy is the most stable institution in India across centuries at this point, right? It’s not surprising that only, say, 5% of people get married outside of their caste. Or is it that our structural transformation has led to such a large-size informal market, whether it’s for goods or services or labor, that it makes everyday life very precarious and therefore less likely to break down your kin and caste and faith relationships? That’s really what you rely on as the safety net.

We saw this through COVID. The moment things are precarious in the city, you immediately go back to the village. Or even in the city, you rely on your caste-kin networks. What is going on with India’s structural transformation that you don’t see things going in the direction that standard theory would have predicted?

RUKMINI: I think one of the things that complicates this—and I should of course start by saying that I don’t feel equipped to have an overarching answer, but I feel like I see parts of it. I think one of the things that complicates the answer is the fact that we do see expected structural transformation processes in some areas. One of the most stunning ones is the fertility transition.

It’s progressing along absolutely classic lines as expected, and even the further breakdown within it—whether it is the differences between states or the fact that the gap between Hindu and Muslim fertility is narrowing fast, and Muslim fertility is falling even faster than predicted. Muslim fertility is significantly lower in richer, better-developed southern states.

All of that is progressing on absolutely classic lines. One of the reasons it becomes hard to understand is why is it that there are some aspects of the structural transformation that are going on while there are others that aren’t. When it comes to caste networks and to endogamy, it really becomes something where you can no longer see these as passive processes.

When you can see the very strong instrumentalization and weaponization of so much of this that is going on, you begin to ask yourself the question of whether it’s valuable enough to let go to enough people and whether there is power in having it continue as it is or even sharpened. If there was a mass social movement, then you’d wonder how these processes are militating against each other and what is driving one and when you’d see one come out versus the other, but we’re not seeing that at all. We’re not seeing a sort of broad-based pan-national movement against any of this.

If no one sets these norms, then I think that’s the part at which I really begin to feel that these cannot be expected to be natural processes, because that norm-setting is simply not happening in India at all. In the book, I talk a little bit about a paper by Payal Hathi looking at responses in surveys to questions on interracial relationships in the U.S. versus intercaste relationships and intercaste marriage in India. Among the various things, why is it that it’s become virtually socially unacceptable to profess views against interracial marriage in the U.S.?

While this in no way means that racism has ended, it does mean that that norm-setting around responses at least in surveys has happened. Then, to reduce all of that to expected or natural or structural processes would erase the entire civil rights movement. This happened because people worked at it, and I can’t look around me right now and feel like there’s a large core of people working around these processes right now. We don’t have a civil rights movement around simply treating Muslims as equal citizens. I can’t even say that norm-setting has happened from the very top right now. In the absence of that, imagining structural processes as a sort of unstoppable force becomes difficult.

India’s Asymmetric Structural Transformation

SHRUTI: When I spoke with Alice Evans about this—and this is, again, you talked about how fertility has been very much in line with what the long-run trends of economic growth and structural transformation are—but female labor force participation is neither in line with economic growth and structural transformation, nor is it in line with fertility dropping.

It’s this bizarre thing that’s happening in India, where women aren’t going to work. She talked about many of the things that you find in the book. One is caste endogamy, a big one; sexual purity, we’re constantly policing women; there’s a lot of female seclusion, which I think comes out most clearly in your healthcare chapter, this female seclusion and being limited to the home and without any—requiring permission, so to speak, to leave the domain of the home.

One of her answers was that what India will need essentially is just—the compensating differential or the wage differential to pull women out of the home has to be higher. But the structural transformation mechanics are still sound, and they hold. I’m simplifying a lot of what she discussed, but that’s the Cliffs Notes version of this. Do you see that happening in the world that we live in at all?

I don’t see higher wages for women. I don’t even think you can legislate or norms-set our way out of this. Because I’ve written about this with Alex Tabarrok: State capacity is so weak that this is going to be one more zombie legislation that’s just never going to end up being enforced. At worst, we’ll criminalize something, and some poor minority businessman will go to jail for it. How does one think about this, specifically in the domain of female labor force participation or women’s access to healthcare?

RUKMINI: Of course, I’m not the first person to think about this, but trying to figure out how much, to what extent social vis-à-vis economic forces play a role in a process is obviously an interesting thing to try and tease out.

I think I sometimes find myself at odds with some of the emerging feminist labor economists writing around the issue, which is the argument that the jobs are not there; that it isn’t just a question of women aren’t available to work, it’s also a question of jobs not being there. I find it very hard to separate this from the issue of what is acceptable to see women doing, because these are not jobs in some sort of antiseptic description. This is what is acceptable for a woman to be seen doing.

I too cannot see wages being what really pulls us out of this. But I will say that I find the female labor force participation statistics around it less surprising than perhaps some others, because it fits in exactly with some of these other processes. As we said, the picture that you begin to draw together starting right from things like, as you mentioned, needing permission to go to a health center.

I think sometimes understanding the huge break and the effort and the big moment that a woman going out of the house to do paid work is, is not something that should be underplayed or underestimated. Because then that takes it into the world of a natural economic process, which I don’t think it is. The idea is pretty much, if we’ve set the norm at going to high school or even to college until you get married, we don’t see that norm-setting around leaving the home and going out to do paid work. We’ve not seen any change in the norm-setting around shared domestic responsibilities, around child care.

To me, it doesn’t feel like that’s moved an inch at all. The fact that Ashwini Deshpande’s work with the CMIE data, for example, shows one of my favorite data points, which is that unemployed men spend fewer hours in domestic work than employed women. That just shows you how far we’ve gone from any sort of conversation in terms of all of this sharing. Even if I just look at my own life, if the opportunity for child care and some sort of shared domestic responsibilities is not made available, the world of paid work is immediately out of reach.

All of these are processes bound together within the structure of a patriarchal household. I absolutely cannot see higher wages being the way out of it. Of course, I know that what I feel about it does not invalidate Alice Evans’ considered positions on it. But from talking to women about what brings them in and out of the labor force, the jobs that they would want, I haven’t exactly seen . . . my sense isn’t that if what I want is a really well-paying job, and that would make it much easier for my father-in-law to permit me to leave the kids with a neighbor or someone and go out to do that work . . . yes.

Democracy in Data

SHRUTI: I want to get a little bit into your discussion about how Indians vote and how little we know about it. Now, our constituencies are basically the largest in the world, in terms of the ratio between the representative to the constituents. On average—this is another crazy thing about India—every constituency is not equally sized or apportioned. Thanks to delimitation, we have some constituencies where you almost have double the number of voters for an MP [Member of Parliament] versus some other southern states.

Let’s say on average, an MP in India represents about 2.5 million voters, and this is about three times the number in the U.S., I believe. The U.S. is second in line. The U.S. also has this problem of underrepresentation. Is our data, and therefore the insights on the Indian voter, terrible because our constituencies are too big and we are not thinking about that kind of fine, granular representation that you actually need to study political processes and political movements?

RUKMINI: It’s certainly a problem in that the sheer heterogeneity of the electorate that you attempt to sample makes sampling so difficult and so fraught and so expensive as well in India. But I would also say that I feel like the quality of questioning is extremely poor, and this is because all of the questioning happens around election, the moment of an election only, and is driven entirely by outcome prediction demands.

We know next to nothing about voter beliefs and motivations and their relationship with politics outside of their relationship with an election outcome alone. Because of that, I think we end up with all the wrong answers. Because we are asking these questions purely at the time of an election when all of the various parts of the voters’ matrix that matter to her, she has decided which of those levers to push because it matters most. That’s probably at the top of her mind; then that is her response for her voter motivations.

That just means that we’ve missed all the rest of it, which perhaps two years before an election, she would’ve been much happier to talk about and much more able to talk about all elements of it. Yes, I’d say I don’t envy people attempting electoral and voting surveying in India. I don’t think it’s easy; I think it’s very easy to get it wrong. And it seems extremely expensive to me, which if it doesn’t come from a nonprofit model, it also has very low returns.

Essentially what we’ve turned all of this election polling into is cricket scores. That’s all you want to know. In fact, polling agencies complain to me that people don’t even want to know the vote share, which is something that you’re more likely to get some reasonable estimate of. They only want seats. They don’t even want that vote share estimate. They don’t want confidence intervals, none of that.

Yes, I think by allowing this to be handed over completely to the media, which is a problem—the media almost solely commissions and pays for election-related opinion polling in India—we have allowed research priorities to be dictated by what primetime news wants. All of that explains why these questions are so poor. Yes, when you think about the size of a constituency, you sometimes wonder whether a single constituency would perhaps tell you much more than a survey that attempts to do a representative sample of all constituencies across the country.

SHRUTI: The fact that the largest democracy in the world, that there isn’t a market for political data, means that there’s either a demand-side problem or a supply-side problem.

RUKMINI: I do think there’s a market for it, but I think that that market is being cornered by political parties who do pay well for reasonably good polling of different sorts, but it’s not public.

SHRUTI: Yes, it’s vertical integration, right?

RUKMINI: Right.

SHRUTI: They’re using it completely for their private use. I was thinking in terms of public data, I see two problems in this market developing and becoming thicker. One, of course, we have these ridiculous limits on campaign spending, which nobody is adhering to. But that means that a lot of the spending happens in cash, which means that a lot of the pollsters, even local—you can never be quite public about the transaction taking place. The data is exchanging hands between the political entity and the people who are giving this information.

The other problem I think is related to the media market. In India, the way we’ve set up licensing of broadcasting means that if you’re a news channel, you have to pay a higher fee for licensing than even regular channels playing devotional music or something like that. Because you have to pay this fee, you need to address the largest number of people or bottom-feed, as they say, to the lowest common denominator. You can’t have subscription TV or a subscription news channel. You certainly can’t have market segmentation. I think the only segmentation we have is regular news and financial news or business news.

Other than that, we haven’t done much in that space. I can think of these two reasons why the market has not really developed as a public market where you can have proprietary data that is bought and sold and subscribed to. What are some of the other reasons? Is it because the election commission has banned immediate release of exit polls? What is happening in the broader institutional arrangement in India that we can’t get the data that we really need, even though this should be available at this point?

RUKMINI: I do think that this is actually a good point to consider the role of publicly funded and nonprofit opinion polling. If you look at the Pew model, for example, I do think that most countries would have a large publicly funded or nonprofit reliable surveyor. The fact that we are having to depend entirely on privately owned television channels seems like a terrible way to go about things.

I also feel like the market for this has not—there is a certain lack of maturity to it, which is that we have gotten to the outcome prediction market. But whether this should even be the same as the market that is there for understanding politics or voter motivations is not apparent to me.

In fact, I’d argue that they should probably be two different things. Lokniti CSDS, which is as close as you’d get to an academic institution that does surveys around elections and around voter behavior, should ideally—they do say that they don’t want to be getting into seat prediction stuff. But ideally, this should be their area: understanding the relationship of the citizen with the state and the citizen with political parties, and this betting market situation that happens around elections, which, to be honest, has a lot to do with political parties and money as well.

A lot of these surveys are paid for in proxy by political parties. Sometimes, it is with an explicit understanding with the channel that it’ll help the political party. Or it is a cross-subsidization model, that there’s no way that the channel can pay that much, but the political party has already commissioned it, and the channel feels no obligation to say who has actually funded it.

We, as viewers, apparently don’t feel any obligation to demand that information, to know who has funded all of this either. There is a long way to go in terms of maturity of these surveys themselves. Honestly, I think there’s a conversation to be had around public funding of elections, not just public funding of opinion polling.

SHRUTI: On the public funding nature of this, is India just too big for that? I always worry about that. I mean, we understand that the scale in India for everything is kind of really outsized, but where are the incentives for someone to study border states in the north and northeast in the same way as studying the coastal states in the south and southwest?

Whether it’s distribution of population, whether it’s the end goal of the people using it, the idea that there is this pan-India anything and that someone would be interested in a pan-India anything, that always strikes me as a bit of a pipe dream. When private channels and nonprofits do engage in this—you must have seen this over and over again—they’ll pick the eight biggest states. Those states are already overrepresented in any kind of data that we are looking at. Is this just a too-big-to-tackle kind of problem?

RUKMINI: I think this is where academia needs to and is, in some ways, stepping in. There is a space for a combination of large corporate trust bringing in money and academia doing this work. For example, the Azim Premji University out of Bangalore—a couple of researchers there have done good work using Nagaland as an excellent example to understand how much of our surveying and counting in border states we get appallingly wrong. I think that work on its own—I think about it every time I write about a “nationally representative survey” that will have separate data for each of the northeastern states with the asterisk next to them, and at the bottom will say that the sample size is too small to say anything about it.

I think I won’t be as blasé anymore about using those numbers. But then I suppose the downside to it is, if you’re coming to it with an attitude of, let’s use the data that we have and get to work with it, the downside is that then if you leave out all the states with sample sizes that are too small to say anything, you end up with those same eight states that you already know too much about. Perhaps not one person or one enterprise is going to tell the entire pan-Indian story, but if there are more pieces of research like the Nagaland story, at least they will bring some life to that asterisk so you know what it is that you’re missing.

The Miniaturization of Identity

SHRUTI: I agree with you there that we need to go local. I would say actually even Nagaland is almost too big as a size to tackle. The more I read books like yours, the more I realize that what we are lacking is context, and the only way to get context is by diving deep. You are able to bring some of that because you have been at national newspapers, traveling all over the country to cover very specific stories. You’ve gone to institutions in different states, spoken to people in different states, taken different modes of transport. But even that basic level of context is usually lacking in virtually anyone who’s dealing with something like this.

My favorite example was when you’re talking about why someone might not vote for the pan-India incumbent party. They may be as nationalist or as much in support of authoritarianism or something like that, but their grouse might be linguistic politics as opposed to religious politics or cow-eating or whether they should build a temple in the disputed Babri site and things like that. That just brings this into sharp focus.

Amartya Sen coined this term “miniaturization of identity.” You’re recognized only by your caste or your gender or your religion or something like that. Whereas we’re all fairly complex, and there are different aspects that pop out depending on what kind of decision-making is in play. When it comes to marriage, we know what trumps pretty much everything else.

When it comes to consumption of Maggi noodles, or when it comes to going into the voting booth in a Panchayati Raj election or even a national election, a different face comes out. Is this just a general problem of anyone working with data—there is going to necessarily be this miniaturization because you are never going to quite get to it? Are there ways around it? Does it have to do with collection? Does it have to do with dissemination? Does it have to do with analysis or context-setting? Can you walk me through the stage at which this problem becomes salient? Or is it at each level of the problem?

RUKMINI: No, it’s a great way of phrasing it, and it’s one of my abiding disappointments with the practice of journalism, which is this Amartya Sen concept of miniaturizing. I sometimes refer to it as flattening. I feel like we’re flattening people on the pages of our newspapers every day—and especially around political reporting, which has become so formulaic. Each person has to represent their caste group or their gender.

I don’t actually know whether data solves this problem or makes it harder. I would like to think that good, representative data that is sufficiently granular can problematize some of these flat narratives. Then you don’t get to say that people in Tamil Nadu vote for the party that gives you the most money because a well-done opinion poll has shown you people’s deeply held beliefs about things like federalism or about their language or about social justice.

I would hope that it does do that. I think it’s a very difficult question for journalists, in particular, to answer about the comparative weight to place on narratives and stories versus data. Because, again, each person you choose to speak to complicates the data in their own separate way. The conceit of data is supposed to be that you should not be looking at those individual stories, because what you’re trying to tell is a story that’s supposed to be zoomed out away from all of them. You know you can’t do that in good faith because you know that this particular individual that you’re speaking to, in some ways, turns on the head all of the data that you’re talking about or that particular narrative.

I encounter it particularly when it comes to voting data, because it is such a complicated decision. It is so sophisticated. It is tied up in so much, from economics to self-respect, to feminist beliefs, to what happens within the household, to all of that. I struggle with the question of how far to go with data versus people’s stories. I think one of the reasons I got into data journalism was because I was tired of this parachuted political reporting around elections, where someone was standing under the village tree and speaking to four people—often all of them men—and giving you their take on what this meant that different groups were voting for.

If I feel that accessing a well-done survey of that area gives me greater insight into people’s multiple motivations, what do I do with that one story of a person who is not captured by that category? I don’t have a good answer to it, but I would say that the things that make you think the hardest and that stay with you the longest are usually stories of people who defy this categorization. We can’t all do this in our day jobs, but at least attempting—or even if it’s just for yourself for a flight of whimsy—to talk to people and even record people who don’t fit into these categories is also a good reminder of the fact that the world isn’t as neat as a survey you’re looking at sometimes might make you feel it is.

SHRUTI: Yes. Journalism has two additional biases. One is what Schelling used to call the statistical life versus the human life, the value of that. You know you are going to associate so much more with the story of that young child or the old lady who lost her eyesight or something like that. There’s a human interest element to it. You know what’s going to be read.

The second problem is the bias in favor of outliers. The classic rule of journalism is you don’t want the “dog bites man” story, you want the “man bites dog” story, which means you want more and more outrageous statistics or the same statistic in a more outrageous context, or without context as the case may be, to be the thing which is in the news.

In some sense, the way we think about newspapers and the quotidian nature, and how it’s fighting for attention with 17 other things, itself makes data journalism a problem. It’s just fraught—we’re trying to do it systematically and consistently—unless there is a greater effort to do it systematically and consistently, right?

RUKMINI: Right. I think this is something in all of its glory and dimensions that we’ve seen in the last two years. We’ve seen all of this outlierism and this search for the most outlandish take on the same set of things. Even the fact that we’ve seen the decoupling or divorce from denominators, which has been deployed across the board by governments and by the media as and when needed. Just that one very indicator has shown all the things that people have attempted to do through data and through data journalism—these competing narratives and cherry picking and manipulation and all of that that’s happened in the last couple of years.

I think data journalism, in particular, has done great good in the last two years, but we’ve seen it for all of its flaws. If there was a particular model of fixing it, I’m not sure that that’s emerged. There is the conceit that the best is what floated to the top, and what you remember are John Burn-Murdoch charts in the FT rather than things without a properly marked axis on the front page of, I don’t know, The Sun. There is that hope that the best is what floated to the top, but we’ve seen all of this in the last two years.

COVID Coverage in India

SHRUTI: Speaking of the last two years, I wanted to talk to you about COVID. There are so many things to talk about COVID and data. You’ve covered it beautifully in your podcast series “The Moving Curve,” where you did pretty much a daily digest to walk us through what was going on, which was incredibly useful. You even won one of our Mercatus Emergent Ventures prizes for it.

There are multiple parts to my COVID data question. One is, it’s lovely that you mentioned what pops up in the Financial Times, the daily graph in the FT or The Economist or something like that. One problem in India is just that the kind of data we’re looking for doesn’t exist. It’s not that it’s a current conspiracy. There were a lot of people who were like, “There’s this conspiracy to hide the number of deaths,” and things like that. I’m sure that narrative serves some people, but the fact of the matter is we’ve never been good about registering births or deaths. We don’t do this well for tuberculosis. [chuckles] We don’t do this well for diabetes.

I think maybe the best death statistics we have is for infant mortality. There is some effort made to collect that data, but why is Indian infrastructure for death estimates so bad, even beyond COVID? And then how did that play out during the pandemic? And what do we know or what do we not know about it?

RUKMINI: I do want to make a distinction between things that have this legacy of failure and institutional failings, and things that have been different breaks from the past in the last couple of years. Death registration, while poor, has improved significantly to the point that we can now say that over 90% or 92% of deaths, we estimate, are now registered. Why this has been poor in the past, I’m sure one of the reasons—and particularly economists must ask—is what is the incentive to register a death, particularly when there isn’t property among poorer households and in rural areas? I do think this is something that the government has pursued hard, and it’s got significantly better.

Another thing that’s got significantly better is the digitization of these records. It was a laborious process, and now almost at least every state has it online every day. Of course, it’s hidden from the public, but it exists. The fact that it has got to the stage is a huge improvement and has happened in the last couple of years.

Where things are much more problematic is in medical certification and ascribing cause of death. Most of this has to do with how people access healthcare, which is, again, believing that the way that disease and death progresses in our households is the way that it does in the rest of the country. This is what would lead people to be surprised by why the numbers are in the state that they are.

A huge number of people die at home without medical attention and with nobody in the family having really a very clear idea of what happened, beyond perhaps having had a fever in the case of all of the vector-borne and infectious diseases that go on during the monsoon, or at most a heart attack or a stroke or a certain event of that sort.

We need to think about this as a spectrum to be able to access healthcare. And then the things that happen when you get better access to healthcare and better hospitalization, medical certification of death will be an outcome from that. It’s not something that’s going to happen on its own. It’s going to happen only when those people are making it to hospitals at all.

Yes, we’ve had a systematic problem about ascribing cause of death. We’ve had this huge issue with malaria in the past as well. This isn’t a COVID problem alone. But then there are a couple of things that have been different in the last two years when it’s come to counting COVID deaths. One is that India decided to adopt, even if not on paper but definitely in practice, a much more stringent definition of COVID deaths than what the WHO, or even the India’s national guidelines, recommended. When it became very clear, in the first wave, that this was happening, there was no reinforcement of this fact that, “Listen, let’s go for the most lenient definition as possible so that we are not missing deaths.” There was not that signal at all. That did happen.

Then, secondly, the question of excess mortality and missed deaths, which is something that countries the world over accept happened, has become something that India seems unable to accept, that there are deaths that have been missed at all. In fact, I feel accepting these excess mortality estimates would put India in a very average place. It would be at par with anywhere else in the world. Once you’ve decided to make yourselves look exceptional as a country that, for inexplicable reasons, had much lower deaths than anywhere else in the world, you paint yourself into a corner from which you become unable to accept any excess mortality estimates.

I don’t think that there is willful suppression of civil registration data. It typically has always been released with a two-year delay. These are not nimble systems, and they’re not going to turn around and produce their data immediately. But again, there hasn’t been any signal from the top that let’s put these numbers out sooner because journalists anyway can access them, and it does say something about a rise in mortality, whether from COVID or from related causes. Let’s analyze it, let’s throw it out, let other people look at it as well.

Then there has been some amount of active suppression. For example, the National Health Management Information System, which is a dataset that’s existed from the early 2000s—once it began to be used by me and other people to look at both disruptions in health services on account of COVID as well as excess mortality, it was just simply pulled offline. That goes into the realm of active suppression, then.

SHRUTI: There’s also unintended suppression by individuals, not institutions, like people who don’t want a COVID death certificate so that they can actually cremate their family members without additional COVID restrictions. Or people who don’t want a COVID certificate because then it’s going to take three more days for the hospital to release the body because they need to correctly categorize it. The incentives on all sides are quite weak and not good enough to check a system which may have been dialed to too stringent to begin with. But there’s also no demand to make it less stringent because people can access the loopholes then and get out of it.

RUKMINI: Yes. The government’s argument is that since compensation was announced, it increases incentives for reporting and as a result would result in over-reporting. I suppose it’s something that can be argued from all sides, and without excess mortality estimates, it’s hard to know how much each explanation accounts for.

Data on Excess Mortality

SHRUTI: Now, what is a good way to think about excess mortality? We know some people, there are at least three papers [P. Jha, C. Tumbe et al.; Anup Malani et al.; A. Anand, J. Sandefur and A. Subramanian] that I remember reading in the recent past. All of them place excess mortality at about the same range. What is a good way to think about excess mortality? What is the range that you have in mind? Is this going to increase or decrease with the Omicron wave? What’s going on in the excess mortality zone in India?

RUKMINI: I think the three papers, broadly within them, would put excess mortality in the range of three and a half up to six million from January 2020 to, say, July 2021 or so. I don’t have a scientific explanation of what seems right and what doesn’t, but I largely follow along with the reasoning, and it makes sense to me. I think there are elements and there are nuances within this, which will be much harder to get into without civil registration system data on the whole. First of all, there are big differences between states that we had to extrapolate for right now.

There’s also the question of improvements in civil registration system over time, though I don’t think that pandemic years were particularly ones in which administrative systems perform to the best of their abilities. These seem like fair estimates to me because I’ve seen the underlying data. I think they’ve been argued in a fairly fair fashion, with the acknowledgment that this isn’t great data. I’m perfectly happy with it being estimates of what excess mortality appears to be right now, with corrections that can be made for when better data exists.

If you look at countries around the world, of course, there are countries with better civil registration system data. But they are not all sitting around waiting for perfectly stratified data as well. They are producing crude estimates. The very fact that for South Africa, in this most recent wave, we have data not just on daily cases and daily deaths, but also hospitalizations and excess mortality as a concurrent graph—I don’t think South Africa is trying to argue that this is perfect data. It is data that they feel obliged to make public right now. I’m comfortable with that.

SHRUTI: In some sense, we’ll know in the future the way we estimate for things like tuberculosis and malaria, and we’ll do verbal autopsies. I’m sure post-pandemic, there’s going to be a big effort. I know some people are already working on it. It feels like one of those things where right now we just have bad data, and we got to trust the experts with good estimation techniques and go from there, and hopefully produce better data. Do you see that move happening in the immediate future, the quest for wanting to know? Or is the problem just one of intent: Nobody really wants to know these numbers, and nobody really cares?

The government doesn’t want to know because then they’ll have to pay more compensation or admit that they got it horribly wrong, or our healthcare infrastructure is weak. In the scheme of priorities for individual voters, better data is so far down that list. Nobody really cares. How do we think about this going in the future? Is it just going to be private and philanthropic efforts to get verbal autopsies, and a paper will come out five years from now?

RUKMINI: There’s two things. One is that I don’t think it’s true that people don’t care about data. If you phrase it in this way, they might not care about it. It might not swing elections, but it’s very common to have conversations in India—where you talk to people about the second wave in particular because it was really a mass traumatic event that swept the entire country—where people will talk to you about how they believe the numbers were wrong, and that it was wrong that the government suppressed them. I think this is a very widely held view.

If there was a particular government that said, “We’re going to do this big household survey, and you come out and tell us if you’ve had COVID or COVID-like death,” I don’t think this is something that wouldn’t be supported by large numbers of people. I do think it is an emotive issue that people do care about. There’s a lot of data that people don’t care about, but this is one of those things that I do feel—just because of the sheer scale. It’s hard to meet an Indian person who did not lose someone in the second wave.

I actually find that that is one of the reasons why the excess mortality reporting was in some ways able to breach the bubble that a lot of my reporting does not sometimes breach, just because of the sheer scale of it. Even people who might not be aligned with your views or might not be reading Dainik Bhaskar because they don’t feel that it’s sufficiently pro-Modi or pro-Yogi Adityanath, did feel in Madhya Pradesh, for example, that everybody knew someone in their native village who had died from COVID and were not seen in the numbers. And they felt that this reporting was honoring what they were seeing.

I also don’t think that the government does not care, but I do think that the government—by making its approach seem beyond reproach—has painted itself into a corner. Plus, this is a government that has shown itself to be very prickly about criticism, and quick to deploy a spin to make numbers look like something else. I do think we are going to have excess mortality estimates from the government at some point, but I think it’s going to be very far from accepting other broad estimates.

Who this is going to come from—again, I wouldn’t necessarily put it past the government to come up with some sort of surveying system. They might, and I do think that it’s possible that some state governments might choose to do this, too. I think some phone surveying methods have shown promise. This really seems like something that should be fairly easy to conduct on the phone. I think we will see some more academic surveying around this. It’s going to be interesting to see what the government decides to do. It’s likely that they will come up with some sort of survey.

Again, I’ll say that I think verbal autopsies are going to be pretty challenging in this context, because of just how ubiquitous these symptoms are. I speak sometimes to doctors who work in rural areas, for example, and it’s a verbal autopsy nightmare, the typical symptoms of COVID. Many of these doctors are actually quite critical of Global Burden of Disease or IHME estimates as well, because they say that they themselves while treating a patient would find it so hard to sometimes say whether it’s typhus or malaria or dengue. To think that someone is able to speak months later to family members and get an accurate reading of which vector-borne monsoon disease-led death it was is going to be challenging.

SHRUTI: I think even there, just with the current data, it would be easier to get an excess mortality estimate than it would be to get a precise cause of death estimate. Excess mortality is useful even with COVID because let’s say the person didn’t die because they got COVID. It could be because the healthcare infrastructure collapsed so much that they couldn’t get medical care in time.

Now, that’s not a COVID death, but it is a death which is largely related to the pandemic and must be attributed to the death statistics that we’re looking at during the pandemic. I think there’s some value in that. I’m sure there’ll be people who try and triangulate a little bit better. You’re right. If we are not counting this in real time through hospital infrastructure and data, I think there is a really big problem there.

RUKMINI: Right. It was probably more useful to come up with close to actual estimates of missed COVID deaths during the first or even the second wave where there were more questions around what you should be doing. Was something going wrong in the treatment protocols? Did we need to think about vaccination? All of those ships have now long sailed. The amount that we alone in India can derive as knowledge around COVID mortality, that window for it has really closed.

We’ve got the information we want from the rest of the world that had better data. At best, we can come to terms with the loss of human life in the last couple of years. If we are in a moment of being more introspective, we can think about how far our health system is from where we want it to be.

Research Bias

SHRUTI: You know, the first thing I realized when I read the book is that I’m an outlier. In every chapter in the book—and I think you are too, right? And you recognize that. Whether it is our political ideas or our liberal ideas or the way we think about our consumption habits, healthcare access, our income and wealth, most importantly. The most traditional thing about us might be that we are married [chuckles]. Other than that, I think we are really outliers in the scheme of Indian women in our age cohort. Now, when you think about putting together a set of themes of looking at India around data, given that you are an outlier, how do you know you’ve chosen the right questions?

RUKMINI: I think starting from a place of recognizing not just that you are an outlier, but that you are an outlier with outsized importance in media, in public opinion, and it’s an outlier position of enormous privilege rather than the opposite.

I think recognizing that is important, and also coming at it with the feeling of wanting to not continue talking within this echo chamber. Because one of the points that you mentioned is about holding liberal values, and I think it was particularly important to me to bring out what opinion polls and surveys show us about just how common it is in India to have casteist and Islamophobic, authoritarian, fascist beliefs. Because this is a sort of comforting narrative that liberals tend to hold themselves, which is that sure, there’s the occasional fringe that displays this sort of behavior, but that the vast majority of Indians don’t behave that way.

I think recognizing that perhaps all of my readers for this book as well will be in this outlier segment because it is a particularly privileged group of people. Perhaps this book will continue to talk to that set of people, but I think recognizing that we are all talking to each other and coming up with these narratives that we all then support and conform to and then believe to be the truth is important.

Then coming at it with questions that seek to encompass more of what the broader reality appears to be from the numbers then becomes important. I was talking to someone else today, and they were talking to me about thinking about the poor, the non-poor, the middle class and the rich in India. How should we think about a middle class, if we see from the numbers that they are so small and there’s such little resemblance to what political science imagines a middle class looks like if you look at them in numerical terms?

I think increasingly trying to make ourselves more and more irrelevant to public discourse is the direction I would love to go in. I’d like to think about the poor, I’d like to think about those who live precariously, not just about the non-poor. I’d like to think of people who are perhaps just past that, but not within this 1% or 2% of income tax-paying people like us who dominate public policy. I think asking questions that have less and less to do with my lived reality is an important direction that I’d like to go in.

SHRUTI: Are there questions that came about after you started doing the research, like things that would have never quite occurred to you, but once you started looking at how people live, you’re like, “Oh, this is something that would never occur to me in my everyday life, and now I need to actually delve deep into this topic”?

RUKMINI: Yes. Some of it was stuff that’s come up in the course of my reporting over time, but some of it was stuff that I realized that I hadn’t spent enough time thinking about and that I really needed to figure out—I just had vague contours of answers. For example, migration is something that I’ve not looked at systematically before. I’ve read Chinmay Tumbe’s fantastic recent book, and I follow broad data on migration, but I hadn’t really thought hard enough to have opinions and a narrative that I wanted to build around it.

That’s definitely something that was a new direction to strike out for me as well. I think I’ve also kept a lot of data around the economy at a bit of arm’s length. I didn’t, in the past, engage with questions around how to think about consumption expenditure versus national accounts data.

I’m not a person with training in either economics or statistics, and so I didn’t feel that I had the chops to do justice to these discussions. But then I realized, of course, I’m not going to elevate that discussion to a level beyond what has been discussed already in India, but I did think that it was something to tackle head-on. Because I do try to make the point repeatedly in the book that the left and right sometimes speak at cross purposes to each other around data, particularly around the economy.

One of the points of deep contention is around how India calculates consumption, thinks about the GDP, thinks about national income, national accounts, all of that. Those were probably two of the areas furthest out of my comfort zone.

SHRUTI: When I was reading the different chapters—the way you’ve set up the book is each of them tackles one subject. You bring an enormous amount of clarity. I started making some connections in my head, and I wanted to sort of pick your brain on is this going in the right direction? Now, if you think about it, how many calories we consume in a day or how precariously we live may not have that much to do directly with, say, our political views or how we think about religion or caste. They seem like quite separate things, but in India, to me, it seems like it’s all entangled.

The more precarious you are, the more you’re likely to rely on your family, caste and religious networks; the more likely you are to want to live in an area where there are other people like you, which is alluding to some of the ghettoization tendencies that you’ve talked about. How much do you see a big composite picture emerging on—these things just didn’t come out of some long-past dependence. They’re actually deeply entangled with each other. Our access to healthcare, how we treat women, the issue of agency and marital rape and how casteist are we and do we practice untouchability are actually much more closely linked than one would have imagined. Am I overreaching with this?

RUKMINI: No, I don’t think that you’re overreaching, though I would say that where I come down on the connections would differ in some ways. For example, I think one of the things that came out very clearly in the chapters around beliefs and attitudes is that these are not attitudes that become more liberal with more income or better education or even among younger people.

It’s tempting to think about these broad drivers, which I know you see in U.S. political coverage as well. If you think about the big-city, postgraduate voter versus someone with the high school education in the rural Midwest, and you think about how those religious beliefs as well as other beliefs, and then how they add up to voting behavior. This is one key distinction over here, which is that some of these processes that we might have expected to start happening in India are clearly not going to go in these expected directions.

We are no longer able to—we have to really get rid of that sort of belief that over time with better incomes, more education and greater urbanization, that there’ll be this sort of natural progression toward liberal values. We’re seeing precisely the opposite impact. You see that being better educated or being richer can predispose you to more authoritarian beliefs in some cases as well. It’s important to figure that out because when you think of these as natural, organic processes, you stop thinking of the work that needs to be put in to make it happen.

You can’t erase the entire civil rights movement in the U.S. and assume that these were natural processes, that then we can transpose onto India over here. We are so far from having that sort of movement, I can’t even come to a point where I can say that there’s a broad national consensus that Muslims need to be treated as equal citizens. We aren’t even at that point yet. So then to assume that these beliefs will come naturally over time is not the direction that we’re going to go in. But yes, I did begin to see connections over other areas.

One of the things you realize is that you repeat the same cleavages chapter by chapter, and then when you do that, you realize that these are fault lines that exist across the spectrum, across different areas.

Then, yes, you do begin to think of what it’s like to navigate life within these areas. You begin to think about what it is like to be a woman, less likely to access healthcare even though you’re more likely to fall sick, what household bargaining power means, what the world of work means. Then these connections do start coming together. I’d say I’m still a great distance from having a grand theory that can encompass both the economic and the social. But yes, I agree that I did begin to see these connections coming together.

Double Life of Young Indians

SHRUTI: One last big question is, one of the things that came out through the course of reading the book is Indians, especially young Indians—everyone’s leading a bit of a double life. There are things that you are willing to share with your family, and there are things you aren’t. You have this lovely anecdote about someone who’s literally packing two bags for his two lives: one where you have a religious thread and all your regular clothes; the other one has the cell phone with which he speaks with his girlfriend or partner who happens to be of a different caste.

I feel even silly asking this question because how can you know how widespread preference falsification is, and how acute that problem is in particular when it comes to survey data? How bad do you think this is in India? The surveyors, we must point out, almost always tend to be educated, upper-caste, majority-religion men. How does one think about this?

RUKMINI: Yes, it’s one of those things that you feel that you have to be forced to cop to, but you do need to think about it. I feel so possessive about all of my favorite surveys. I feel so sad to have to think about the possibility that they’re missing a lot of things. But yes, it’s one of those things that talking to people immediately raises big questions about the numbers you’re seeing.

I think the example you picked is actually well-chosen, because I would find it very hard to put numbers on the extent of preference falsification for premarital sex, which I do think happens. Because I suspect that the number is on the lower side—what official statistics show—even if not premarital sex, then at least premarital relationships. It is my sense that there is much more of it happening to some extent, even if it’s just texting over Facebook; I think there’s more of it happening than what the data captures.

I suppose the best thing would be to, again, think about triangulating from multiple data sources. We’ve seen this in the case of work done around voter preferences, for example. The same voter will say that caste doesn’t matter to them and that they’re voting for jobs, but then the same set of voters will also—45% of them will say that they want to see an MP of their own caste. We’ve seen surveys where people do this much more effectively, which is to take the question back to you in a different way, and then see how you respond to that.

I think there is scope to ask better questions about things like the practice of untouchability and casteist practices. I think we currently ask a very limited set of questions, and there are so many better ways to be asking these questions. It’s actually disappointing to think that someone had the opportunity to do it and asked a not very great question, ultimately. I think allowing people to self-categorize more is obviously a coding nightmare post-surveys, but is going to be something that potentially presents interesting answers. The IHDS [India Human Development Survey], for example, lets women define what an intercaste marriage is themselves, which is an interesting way to think about more questions that you could ask that way.

I would say that I also want to be cautious about the fact that when I suspect that there is falsification or not admitting to things going on in surveys, it’s probably ones in which I feel unconvinced of the response because of my own personally held belief. I feel by looking at the microsection of the world that I see that there is a lot of premarital relationships, but if this isn’t borne out in the data, I don’t think I should necessarily immediately make that leap to question the data. Because then why am I not questioning the data about all these other things that I’m perfectly happy to accept? Yes, hard to quantify, but important to work on.

SHRUTI: I also think maybe we need to look at data that is not survey data. There was this interesting work that was done by Seth Stephens-Davidowitz, the stuff that he did using the big data of Google. This was anonymized Google searches in a book called “Everybody Lies.” I know this method and the book is fraught with problems and lots of benefits.

What he really found was that people are asking questions through Google that they would never ask another human being, because of embarrassment or preference falsification. It could be political, it could be family. Of course, the most outrageous and the least surprising questions have to do with porn.

Now, when I think of a method like that—one is, I think it could be a great accompaniment or a complement to what is going on. If you’re thinking about premarital sex or questions to do with sexual intimacy, maybe Google data is a good place to look. On the other hand, when you’re thinking about caste and religion, the way we grew up in India and the way we absorb this through osmosis—a three-year-old knows this stuff.

You’re never going to Google a caste question in a way that truly reveals your preference. I can’t imagine too many people Googling and saying, “Is the unwillingness to allow such and such person to cook food in my kitchen untouchability?” Because they would just know it in their bones that this is acceptable, or it’s not acceptable.

RUKMINI: What you do see, for example, is that the autocomplete feature in Google throws up the second or the third option for any name in India as, “What is someone’s caste?” People are questioning, if not the practice of it themselves, the centrality of it to their lives and to their interest in any other human.

SHRUTI: Absolutely. You can get the broad brushstrokes of how people think about sex or people think about caste, but something more nuanced like, “Are they really practicing untouchability, or a nuanced version of untouchability?” Not the old-school, “your shadow cannot fall on my front yard” sort of thing but the more nuanced forms of discrimination we see. I don’t really imagine people are Googling this stuff before they actually do it.

I do see a bit of a revolution happening now with mobility data, for instance. So much of that has been used during COVID for understanding if lockdowns were successful. My co-authors and I wrote a paper on this, on how you technically had the same de jure rule during the first lockdown in 2020 (April to June in each state), but the de facto mobility increases and decreases are completely different in each state. And they don’t have that much to do with COVID spread.

Whether it’s congregation data, whether it’s mobility data, people hanging around hospitals or crematoria, these are things that are now being used to also add to the excess death picture or something like that. I do think there’s some role that big data can play, but I think the role is mostly triangulation. It’s not a good substitute for survey data, is the way I think about it; it’s orthogonal or a complement to it.

RUKMINI: Yes, it is one of my big regrets around the book, which is that I didn’t have any access to Google, Facebook—any of the data. I read the book called “Dataclysm” by one of the founders of OkCupid, and I was just thinking how much I would’ve enjoyed looking at Bharat Matrimony or Shaadi.com anonymized data to say these things. But I haven’t ever got very far with any of these big tech companies in getting any anonymized access, and when the data is typically released, it’s in their curated fashion and very selectively.

Increasingly, there are questions that are not going to be answered by the traditional data sources that I look at in the book, for example. And I would like very much to be able to turn to Facebook. I think there’s a lot that we possibly could say about premarital relationships that could potentially emerge from Facebook, or in the past even from Orkut, which was really one of the first stranger-interaction websites in the new internet in India.

SHRUTI: I’m so glad you said Orkut and not ICQ or something, which is really going to date us.

RUKMINI: [laughs] Yes, I think I edited myself there.

SHRUTI: I’ve ruined it now. Now people know exactly what generation we are.

RUKMINI: I could even make the sound of the ICQ pop-up if called upon.

SHRUTI: We shall not call upon you. We should pretend to be young, cool, woke millennials for the purpose of this conversation. Now, when I go to India, I have fully embraced my aunty-dom. I know the demographics, I have seen the numbers. I am older than almost two-thirds of Indians, maybe a little bit less. You know what, I just got to own up to it. I’m comfortable with this. Let’s just move on.

RUKMINI: Yes. You regularly now speak to people born after 2000. I remember being shocked by people born after 1990.

SHRUTI: Yes. Well, we’re doing something called The 1991 Project, and the entire team is born after 1991.

RUKMINI: How was that even allowed?

SHRUTI: I know. These fertility advances didn’t go far enough. How did we allow these people to be born?

Writing Process

SHRUTI: Before I let you go, I have two questions. What’s your regular writing process and the writing process for the book?

RUKMINI: I think my regular writing process tends to be driven a lot by reporting. It’s very much structured around the questions I ask people, and then the story—I have only a very vague working hypothesis to start with, and the stories are structured around that. They’re also quite manically written because I often work on tight daily news deadlines.

For the book, I tried to make sure that I wasn’t writing it like a bunch of newspaper articles. I did try to think of it more in terms of amassing structural knowledge around an issue in one institutional place so that we weren’t exhuming and having these same conversations around some established facts again and again. But while also trying to make sure that I weaved in all of the things that called into question the data or that the data was missing, as well as conversations that I’d had as I went along.

I think I wrote in long bursts with one issue very much in my mind and tried to bring together both my reporting and my reading around it. The fact that I had these structured into ten fairly distinct chapters helped. That really helped clarify the way I wrote. I also wrote very fast, which actually worked. I stayed engaged in exactly what I was talking about right through the whole thing, and there wasn’t that feeling of pulling away from it and finding it hard to pull yourself back into it. I wrote it all in a concentrated burst of work.

The thing that actually happened is I had intended to take time off from my regular journalistic work while I was doing this, but because it happened—most of the writing was in the first half of 2021, ending in May 2021; that is, during the second wave—I ended up doing quite a lot of regular reporting as well. I actually think that was a good thing because it kept my ideas alive. The BBC’s bureau chief in India, Soutik Biswas, once mentioned that he had read somewhere else that when you have a writing block, what it often is, is actually a reporting block. And I really have tried to take those words to heart.

I wouldn’t say I had a lot of blocks, but at every point, if I tried to think about what more I wanted to say or how I wanted to bring things together, I made sure I spoke to someone. And just that process of talking and reporting brought these things together. So in that way I’m lucky that it is a reported book because I’m not alone with my thoughts and ideas. I’m able to constantly be in conversation with someone.

SHRUTI: Do you write every day? Do you listen to music? Do you sit in the same spot? Do you sit where you’re sitting right now?

RUKMINI: No, I don’t have any of these rituals because my day is circumscribed by my kids, who are six and four and were even younger then. So I can’t commit to any of these rituals. But what I did do during the writing of the book is that in three installments, my parents and then my husband and in-laws came in and did a lot of the child care then. I’d go over to write at a friend’s house, and she was doing some work as well then, and she had a big dining table. And we’d both sit quietly and work and drink our coffee early. That’s where most of this got done. This room is just for show.

SHRUTI: My last question before I let you go—you’ve been very productive during the pandemic, so I don’t know if you’re going to have an answer for me. You’ve written a lot of reporting. You’ve written a book, and you’ve managed to raise two kids through all of it. But what did you binge-watch during the pandemic?

RUKMINI: I binge-watched “The Bold Type” at one point, and most recently, I binge-watched “Emily in Paris.”

SHRUTI: I think we all did; you’ve just admitted to it.

RUKMINI: I disliked it. I liked “The Bold Type,” so I don’t feel embarrassed by that. But “Emily in Paris,” I watched it, disliking it, but I continued to do it. But I don’t think I was binge-watching a lot while writing then. I was just so tired by night that I wasn’t watching much then at all.

SHRUTI: Thank you so much for doing this. This is such a pleasure.

RUKMINI: Thank you so much for having me, Shruti.

SHRUTI: Thanks for listening to Ideas of India. If you enjoy this podcast, please help us grow by sharing with like-minded friends. You can connect with me on Twitter @srajagopalan.

In the next episode of Ideas of India, I speak with Rajesh Veeraraghavan about his book, “Patching Development: Information Politics and Social Change in India.”

Submit a Letter to the Editor
Submit your letter
Subscribe to our newsletter