Data Day Care (with Emily Oster)

[music]

00:05 Paul Matzko: Welcome to Building Tomorrow. A podcast about the ways tech and innovation are making the world a better place. I’m your host Paul Matzko. And if there is one phase of the life all but guaranteed to make you realize just how out of your depth, you are, it is parenting. Your broads with dozens of decisions from whether to swaddle, to sleep train, to breastfeed, and so on. And you have to decide what to do now, all while in the sleep-​deprived state. And the consequences of choosing poorly seems so vast. It’s 2:00 AM, your two-​year-​old is struggling to draw a breath, you take him to shower, steam it up to help them breathe but you’re faced with a decision. Do you rush them to the emergency room because it’s actually a life-​threatening complication of a serious illness? Or it’s just a bad case of croup, and you’ll feel stupid for over-​reacting in 15 minutes when it resolves? How do you know? Parenting is a constant charm between sheer delight, mind-​numbing TDM, and existential dread.

01:02 Paul Matzko: Our guest today has a better way that we can approach these kinds of decisions. Dr. Emily Oster is a professor of economics at Brown University, and the author of two New York Times best-​selling books, the latest of which is “Cribsheet: A Data-​Driven Guide to Better, More Relaxed Parenting From Birth to Pre-​school.” Welcome to the show, Emily.

01:21 Emily Oster: Thank you for having me.

01:23 Paul Matzko: So, to start, Emily, can you pick a topic? Walk us through the mental thought process that you encourage parents to go through when encountering a parenting decision.

01:35 Emily Oster: So, let’s seek something simple, like swaddling, which there are people who will tell you slight swaddling is super important and it helps your kids sleep, and then there are people who will tell you, “No, actually you can’t swaddle because it will ruin your kids hips,” or some other thing or it’s not nice to them or something else. So what I suggest parents do is sort of a two-​step thing: First, really look at what the data says and so, in the case of something like swaddling, what I do in the book is really go through, “Does swaddling actually seem to improve sleep? How good is our evidence on that?” And in that case, the answer is yes. And there are evidence that’s actually pretty good, probably because it’s easy to watch the same baby swaddled than not swaddled and to get a sense of why they sleep better, and whether they sleep better. And then once you have that data, and you kind of really know, what are the benefits and what are the risks which, in that case, are basically there are no risks, then you’ll think about making a choice that works for you.

02:40 Emily Oster: Now, with something like swaddling, it’s not super complicated in the sense that once you see the data, it tells you this is something that’s gonna help your kid sleep better and it doesn’t have any downside. So, on the one hand, not like I have to swaddle, on the other hand, it’s probably a good idea. But then there are a bunch of other things in the book, including some much more fraught topics like sleep position. Do you co sleep with your baby? Where I think it’s more, the evidence is more mixed and once you see the data, then you really wanna think about, “Where are my family’s preferences? I have to combine this data with some understanding of what’s important to us and what are the kinds of risks that we are comfortable taking to make a decision that works for us?” And what that means is that for a lot of these things, the same data is not gonna lead families all to the same decisions.

03:28 Paul Matzko: One thing I appreciate about your book is this sense of… Look, we need good data, and better information will help us make better-​informed decisions, but there’s variables that are gonna be different from person to person, like, you mentioned family preferences, and I really appreciated that approach, that there’s not necessarily one the hard and fast rule on some of these issues. There are some where the data is a little more overwhelming, where the downsides are so slim and the upside is so large that it’s hard to imagine it not being a good decision for most people, but that ambiguity, I think was useful after serving my own share of parenting books where there’s lots of the say of the parenting expert. [chuckle] Right?

04:13 Emily Oster: Yeah, yeah, and I think that attitude, there is a right way to do it, and it is this, and it is the right way for everybody is part of what makes some of these parenting stuff so anxiety provoking because people are so sure that they are right, and that they are so right, that it must be the right choice for everybody. And so then, when they come to you, and people come to give you advice, it’s not like, “Hey, this is what work for me. You might try it,” but “This is what you have to do because if you don’t do this, you’re gonna make a terrible choice, and then your baby is gonna suffer.”

04:42 Paul Matzko: Your baby is like one of those Romanian orphanage children, you’re scaring them for life.

04:46 Emily Oster: Exactly, you’re scaring them for life by everything. Basically, everything you do could scar them for life where, if you do the other thing, it also could scar them for life. There’s no winning.

04:54 Paul Matzko: Yeah, it leads to kind of a paralysis to some extent.

04:57 Emily Oster: Yeah, exactly.

04:57 Paul Matzko: Yeah, yeah. Now, I did have a question. Why does it feel like… Maybe I’m wrong, maybe I’m off base with the presumption, but it does feel like information behind a lot of these parenting decisions is relatively inaccessible and not as well-​distributed in the population, as information is the baseline information of other domains. So, is it worse, and why is it worse than, say, learning how to manage your finances?

05:31 Emily Oster: Yeah, I think that there’s a few things, one is that the data is sort of not… The data is often not as good. It’s… There’s kinds of information that we want, the kinds of data that we want here really require people to make different choices with their kids, so we can compare them and see if different things about them. And you could say the same thing about finances to see if it’s a good idea to invest in index funds. We need to see some people who invest in index fund, people who don’t, but there’s actually also some raw truth there, which is just like, “What if you had invested in index funds,” there’s an actual answer to that, whereas, in this case, parents make different choices. It isn’t always so easy to infer from the data what actually is the truth. And I think that the fact that we’re really dealing with behavior and not just the behavioral financial markets, that’s part of what makes these data so challenging.

06:28 Paul Matzko: Okay, now I’m gonna get you to help us think like economists when it comes to parenting, and you’re actually quite good about not over-​lading the book with jargon, but there are a few things I want you to walk us through. So in the book, you talked about thinking through constrained optimization, what is that and how this thinking like an economist about that help someone approach the trade-​offs involved in parenting?

06:55 Emily Oster: So when we do economics and we think about people, our base idea is that people are optimizing. So that they are trying to make choices about their behaviors, or about their purchases that are gonna make them as happy as possible. That’s their goal. But we almost always give our people some constraints. So on the one hand, I want you to be as happy as possible, but you have a budget. So even though the thing that would make you as happy as possible is to own six Gulfstream jets.

07:22 Paul Matzko: Yes, please.

07:22 Emily Oster: Actually, you can’t have that. So really, we’re gonna give you some amount of money and then we’re gonna think about your choices in that. Maybe you’ll only get one jet, or maybe no… Could be no jets. Could be none. So that, this idea that people are making sort of optimal choices under constraints, that’s kind of a key underlying principle of economics. We think about it applied to parenting. I think we sometimes will sort of frame things as how do I optimize? How do I do everything the best? What is the sort of most safe best, I should be both engaging in play with my kids at every moment, reading to them at every moment. Also, I should boil all the water to make their formula. Also, by the way, you shouldn’t make formula, you should just breast feed. Also, you should pump all day at work. Also your kid should sleep in exactly the right way, even though… Even if that means you never sleep.

08:17 Emily Oster: And I think that a lot of that discussion ignores the idea that people face constraints. That’s there’s a limited number of hours in a day. That you can’t actually do all of those things if you literally haven’t slept at all. And so, some of what I talk about in the book is sort of about sort of thinking about, “Okay, there are some constraints.” And you maybe you are not gonna be able to do all of these different things in exactly the way that everybody says that you should do them. And you may have to pick which of these things are most important. And that’s where I think the data can be super helpful because a lot of our recommendations do not come with advice about… With information about how important is it to do one thing versus the other. Once you have that information it’s easier to say, “Okay. With the constrained time, and there’s constrained attention and constrained money that I have, what are the best choices that I could make?”

09:03 Paul Matzko: Okay, so our choices are constrained. We’re all facing trade-​offs. And maybe the choice isn’t between me playing with my kid or that kid watching an iPad. It’s not like one’s replacing the other. I have to get ready for work. My kid has to do something. So well…

09:20 Emily Oster: Yeah, or even something… In that example, it’s like maybe the choice is the only way for me to be able to make dinner is for my kid to watch half an hour of TV because I can’t actually cook with my kid crawling all over me. So, I could order take out and then I could not have them watch the iPad. Or I could make dinner and they could watch the iPad. But you’ve told me both it’s very important to have a healthy home cooked food, and that my kid cannot watch the iPad. I literally cannot do both of those things. And just like… And so that’s it. You have to choose one of those things, and I think we often don’t recognize that you just, you may have to make a choice.

09:54 Paul Matzko: Now, so I get constrained optimization. There’s another term you throw in it later on the book about talking, about thinking like a Bayesian. First it was a frequentist, but what does it mean to make parenting decisions like a Bayesian?

10:11 Emily Oster: So I talk about this in the context of TV. But I think it actually comes up a lot. So there are different ways to incorporate data into what you think is a conclusion about some relationship. So, the example I give is about Sponge Bob and reading. So, let’s say that there’s a new study that comes out that says that watching Sponge Bob Square Pants helps kids learn to read when they’re very young. So let’s say that’s the only evidence we have. There’s only, amazingly, only one study in my imaginary world that has been run on that important question.

10:48 Paul Matzko: It’s very specific that one study.

10:50 Emily Oster: Very specific, very specific. And let’s say it shows big effects. If you’re what we call a frequentist, somebody who basically is gonna use only the data that we have to draw conclusions. You would kind of be forced to conclude that Sponge Bob teaches kids to read. If you’re a Bayesian, you wanna take that data, but combine it with what we call your prior with some baseline belief about the relationship. And in that example, you probably do not think it is very likely that Sponge Bob is going to affect kids’ early reading. So it’s true that when you find some new data that shows that it does, you should move your, what we say, move your prior a little bit. It should change your beliefs a little bit. But since your beliefs are already so strong in the direction of that cannot possibly be true, when you move your beliefs a little bit, you probably still don’t think it’s true.

11:38 Emily Oster: And so, I talk about this in the context of sort of thinking about evidence on screen time, where unfortunately, our evidence is very, the actual evidence that we have is very limited. And I use this as a frame for sort of a how can you conceptualize what is likely to be true? What are your priors about what is likely to be true? And that that may help us think about that choice. And in particular, your prior is probably that watching nine hours of TV a day is not good for your kid. Watching, spending 20 minutes on an iPad three times a week is almost certainly fine. And so, of course, it would be great to have data about those but actually our priors are pretty strong in those cases. And that then, there’s some space in the middle. Like is an hour of TV too much? What about two hours? Where, really, it would be great to have more data because I think that our beliefs are kind of mixed on that. And it’s hard to know what the answer is and so we could benefit more from more data.

[music]

12:46 Paul Matzko: So you’ve written first here about pregnancy and then parenting, and that makes me wonder how that shifted how do you think of human rationality? And here I want… If you could touch on the divide between classical economists and behavioral economists and it’s an easy distinction to overstate, but there can be a distinction between this emphasis on people as either rational or irrational actors. And I know, I think both you and your partner are economists. Your parents were economists. And I think, at least one of you has behavioral economics training. So has this experience of writing these books of parenting changed your view of human rationality?

13:29 Emily Oster: That’s interesting, so just to give a little bit of kind of how I see the background a little bit of this behavioral stuff. When I was in graduate school or really late in college, in the sort of 2001 kind of space, I would go to the…

13:46 Emily Oster: There was a very clear idea. Behavioral economics was something that was very separate from… It was a kind of economics, where we were moving away from the idea that people were totally rational, and it was like a really… There was a behavioral seminar in grad school. And there were people doing that as their field. And I think part of the tremendous success of behavioral economics over the period since then is that I’m not… People do sort of think of it as a field, but not in the same kind of very separate way. So there’s a lot of people whose work sort of touches on behavioral economics, or tries to use ideas from that to think about data, or to interpret the things that we’re seeing in the data. And so I think that’s… I think those distinctions have blurred a little bit. And I think we’ve gotten more comfortable with the idea of sort of thinking about people as non-​rational. And with really trying to focus in on the places where they’re so non-​rational that it really changes the things we see in the data, as opposed to everybody’s a little non-​rational. But sometimes it’s not, sometimes is not quantitatively important.

14:57 Emily Oster: When I think about about parenting, I do think there’s an experience when you’re a parent, which is, it’s kind of inherently not rational. And inherently, just you kind of can’t believe how much you care about this thing. And also how difficult it is to make choices that you… Even choices you kind of know are right, it’s just very difficult to do that when they’re hard. And even things like sort of the investing now for benefits later, I find, they’re actually very difficult in parenting. I think a lot of people do. It’s very easy to be like, “Oh, I’m just gonna given into whining right now.”

15:55 Emily Oster: But I think it also reinforces some of the things… The sort of key principles of economics, like incentives matter. You kind of know, you know it’s easier right now to give into the whining and then deal with it later. That’s kind of like what we’d say like present bias. Like I wanna… I don’t… I’m discounting the future. On the other hand, it’s such a crisp illustration of the idea that your… If you indicate to your… If you provide your kid an incentive to whine by giving into them, they will definitely whine more. So kids are a good illustration of economics, for sure.

16:29 Paul Matzko: For our listeners who haven’t spent a lot of time thinking about the differences between double blind studies and observational studies, et cetera. Can you tease out some of the distinctions that you make in the book when it comes to evaluating the evidence? So you’ve got different kinds of trials, different kinds of studies, and you’ve also got the issue of correlation versus causation. So, unpack some of that for us.

16:57 Emily Oster: Yeah, absolutely. So I think in all of these places, the basically, what the thing we are trying to aim for is an understanding of if I do X will Y happen? Is there a cause or relationship between this thing that I’m gonna do with my kid and some outcome that I care about? And there are different ways you can study that. And so, the kind of best way, the most causal, wonderful way that we all love is some kind of randomized control trial, which would work by taking parents and kind of randomly having some of them do one thing, and some of them do another thing, and then comparing the outcomes. And if you’ve chosen who does what randomly, then any differences in outcomes you can attribute to the treatment. And there are some things where we can study them that way in parenting. There are not a lot of things like that. And I think you could sort of see why that would be. It is because people do not like to experiment with their baby. So there are a lot of cases where all of our evidence is gonna be from what we would call observational data, where people make different choices and then you compare the outcomes for the people who make one choice, the people who make the other choice.

18:12 Emily Oster: And that’s like… There’s those sort of fundamental correlation versus causation problem with those studies, which is that the people who make one choice are different from the people who make another choice. And as the choices get, in some sense, more and more important to people, that may be more and more true. So in something like breastfeeding, the differences between people who choose to breastfeed and people who don’t are enormous. They’re really, really big. Women who breastfeed tend to be more educated, tend to be richer, more likely to be married, more likely to be white. There’s tons of differences between these groups, many of which also relate to the other kinds of outcomes we’re interested in studying. And so one of the things that I think I do much more in this book than maybe would be true if I were not an economist, is to say, “Okay, within that category of observational studies, there are some which are better and some which are worse related to the question of how good a job they do at adjusting for these differences across people.”

19:14 Emily Oster: So in the case of something like breastfeeding, you have some observational studies where we just kind of compare the two groups and maybe we adjust for differences in the education of the mother. Or differences in her age, in her education, and that’s kind of it. And then you have other studies where we compare siblings, where one sibling is breastfed and one sibling is not. And so, it’s true those are both observational studies. They both are not randomized, but it is a lot better if you can compare two kids with the same mom who are different, who one of whom was breasted, one of whom was not, to comparing two kids of totally different moms. How worried are you about differences? Sure, there are probably some differences. Why was one kid breastfed and one was not? That’s a good question, but that’s oftly… That’s much, much closer to the kind of experiment that you wanna run than the just controlling for a couple of things. And so a lot of the kind of work of the book really is within this very broad category of studies. Trying to isolate the ones that do a better job of separating out correlation and causality in places where we simply do not have randomization, and where probably we will never have really good randomized data.

20:33 Paul Matzko: So now that we’ve laid some of the structural backbone of the book, economics-​based thinking, data-​driven thinking behind it, and a few specific questions about some of the content. So… We’re a tech and innovation focused podcast. Is there… If you had to, under coercion, pick one area that you studied while writing the book that you think would be most potentially productive for technological innovation to dramatically improve outcomes… If we threw venture capital money and the smartest engineers in the country and Silicon Valley at it, what would be that area?

21:14 Emily Oster: I harp a little bit in the book on breastfeeding technologies and particularly pumping technology, which is very poor and very annoying. And we’ve had some innovation in hands-​free pumping, but even in things like just storing, cleaning… The mechanisms that go into making that work is incredibly complicated and annoying, and so that is a place where I think we could use more tech. I’m trying to think of what else. Everybody is always sort of talking about different kinds of things about baby sleep, although I’m not sure that technology is so well-​suited to helping your baby sleep. I guess there’s the SNOO, I don’t know.

22:05 Paul Matzko: Yeah.

22:05 Emily Oster: I don’t know how to comment on that. So I’m gonna go with breastfeeding as being a big thing. Oh, and the other thing is, did you know that a quarter of kids don’t poop in the potty?

22:16 Paul Matzko: Really?

22:17 Emily Oster: Coming back to discussion of… Yeah, or…

22:19 Paul Matzko: Huh.

22:19 Emily Oster: Are potty-​trained for urination long before bowel movements. And there’s a name for it. It’s called “stool toileting refusal.” And so, if there’s some technological way to improve kids willingness to use the toilet, that’d be awesome.

[music]

22:49 Paul Matzko: Did you come across a convincing explanation while you were researching these topics for why both infant and maternal mortality rates are so much higher in the US relative to other developed countries?

23:01 Emily Oster: The question of infant mortality is something that I’ve actually worked on in my academic work. And I think there, I’m not sure we have so much of an answer, but I think that one of the things that we sort of pulled out when we looked at the data comparing the US to some places in Europe is that the issues in the US seem to arise in the second… In the kind of period from a month to a year. So we often separate infant mortality into neonatal mortality, so it’s the first 28 days, and post-​neonatal mortality, which is the period from a month to one year. In the US, actually pretty good in the first month, in terms of actually having fairly low neonatal mortality, particularly relative to the birth weight. So actually, the US is a pretty good place to have a low birth weight baby in terms of how they do. But then, in this period from a month to a year, we see huge differences open up between the US and other places, particularly among poor families. So that kind of suggests that there’s something about the support that we’re giving people not in the hospital but at home that must be contributing to this, but that doesn’t give us an answer. And maternal mortality is something that I don’t… I haven’t studied that specifically. My guess is that some of the same demographic issues are gonna crop up. Maternal mortality is high in the US, although still fairly… Still quite rare.

24:31 Paul Matzko: In historical terms, yeah quite rare, yeah.

24:34 Emily Oster: Yeah, yeah.

24:35 Paul Matzko: One more chapter, I’m sure it’s… My guess is the amount of feedback you get for that chapter is disproportionate would be the vaccination chapter in the book. And you note in that discussion, and I should just say for our audience, you’re broadly pro-​vaccination in there, but you note that the odds of someone being anti-​vaccination actually increases with education, which was a bit surprising to me in the sense that this book is predicated on the idea that better information will lead to better decision making, and I suppose education correlates with being better informed…

25:10 Emily Oster: Yes.

25:11 Paul Matzko: Well, maybe about beer pong, at least. I don’t know. But let’s just stipulate it. Education means better information, but yet better-​educated people are making a scientifically poorly informed choice to not vaccinate their children. So what gives?

25:23 Emily Oster: I don’t know. Oh my god, I wish I understood. So I think this is very puzzling, in the sense that almost every other health outcome you look at, obesity, heart disease, strokes, cancer, diabetes… Basically any short, long-​term outcome that you look at, more educated people do better. So more educated people are healthier on every dimension. It’s one of the most robust relationships within the field of health economics, except for this thing, where it seems like, again, education seems to correlate with a smaller rate of vaccination even though this is clearly a good idea. And it’s actually… The demographics, I think, are kind of are mixed in the sense that there are also pockets of poorly educated people who aren’t vaccinated, and so you see sort of this, these kind of two groups or multiple different pockets of who is not vaccinated. And I don’t really understand what the issue is. I think that there are certainly people who believe that vaccines are dangerous and are gonna cause issues, and I’ve talked to some of those people and I don’t really understand… Still, even after those conversations, have a very poor understanding of what the real issues are. And I think without that, it’s hard to think about why those are cropping up more in more educated people.

[music]

27:15 Paul Matzko: So you’re an academic economist, you’re very active in your field, bringing lots of well-​regarded papers for an academic audience. So, why parenting books? It takes time out from the publisher parish regime of academic publishing and while writing two New York Times best sellers is a reward in of itself, that wouldn’t have been I would suggest the most obvious outcome when you started this whole process. So, what compelled you to say… I’m gonna take time out from academic publishing and write these books?

27:50 Emily Oster: Yeah. The first book was really like… I don’t know, I know a passion project is too strong award, but I got pregnant and I was just like, “Oh my goodness, I cannot believe how frustrating this experience is.”

[laughter]

28:05 Emily Oster: And I started doing all of this research really in the service of my own pregnancy and just finding myself basically like every evening and weekend, looking up papers and trying to figure out what was the right pre-​natal testing and going through the statistics. And I think that the book there, the first book really came out of that, and why exactly I decided to write it in a book, I’m not sure. I’ve always really enjoyed the challenge of explaining these kind of things to people who are not professional economists, it’s part of why I like teaching and it’s a kind of writing that I like to do. And so, I felt, I’ll try to do this a little, and then it got into being more than a little. And I think I didn’t think too much at the time about the professional consequences, which was probably a mistake, but okay, such is, such is life. So then, I did have some pretty significant professional consequences that were not positive. The first book. And so then, I wasn’t actually sure about writing a second book, but that’s why it took so long. But eventually, I realized that in spite of the professional downsides, the social value upside was big. So, my most cited paper has 800 citations in Google Scholar. It’s about using selection, it’s like using… It’s about econometric methodology, and that’s great.

29:53 Paul Matzko: That’s great for that.

29:55 Emily Oster: It was a well-​cited paper, a lot of people use it for their papers, whatever. But I’ve sold like… I don’t know, 100,000, a bit more than 100,000 of the first book. And so, the scope of impact that I can have is much different, and although I very much enjoy helping people improve their political economy papers by addressing the selection on unobservables, getting to be the privilege of getting people to have a better experience of pregnancy or childbirth, or raising their kids, that’s a different scale of impact. And I think, ultimately, that was what drew me to writing the second book. Also I actually really, really like writing these books. I just really enjoy it. And so now, I have tenure and I can do the stuff that I like a little bit.

30:45 Paul Matzko: So I mentioned earlier you and your husband, Jesse Shapiro, are both economists, your parents are two prominent economists, Sharon Oster, Ray Fair. How much of a difference did growing up in a house full of economists make in your upbringing? And do you see any similar effects on your kids? And I know this is anecdotes of just you, but yeah.

31:05 Emily Oster: Yeah. I do think that some of what I tried to, I think is useful in these books, is that I tend to think about things like my parenting or things like outside of my job through the lens of economics, and they like economics of the every day is very prominent in my world. And I think that is partly because of my childhood. My parents… It’s not that they were coming home to tell us about economics research, but the ways that they made choices mostly were grounded in basic economics principles. And so, that idea that the answer to why don’t we go grocery shopping is, why do we have our groceries delivered was like… Well, my opportunity cost of time is very high, which was the answer I got from my mom that was like, “Okay, yeah, that’s the answer.” Yeah, of course. Obviously, it makes a tremendous amount of sense. And so, I think that did influence. They’ve made me interested in economics, and so on, but I think it actually probably influences the book stuff even more than my job. In terms of my own kids, I don’t think they are that interested and I can’t…

[laughter]

32:19 Emily Oster: We do talk about economics a little bit, but not actually that much. And I will say my daughter’s way more interested in biology kind of stuff than she is… And than she is in economics. I try to tell her things about economics research and she’s like, “Yeah, that’s enough, thanks.” She’s just like, “I don’t wanna hear about this.” So, we’ll see.

32:40 Paul Matzko: Well, Emily, thank you so much for your time. Really appreciate you coming on the show, and for all of our listeners, until next week. Be well.

32:50 S?: Thanks for listening. Building Tomorrow is produced by Tess Terrible. If you enjoy Building Tomorrow, please subscribe to us on iTunes, or wherever you get your podcasts. If you’d like to learn more about libertarianism, find this on the web at www​.lib​er​tar​i​an​ism​.org.