Emily Fair Oster is an American economist and bestselling author of “Expecting Better” and “Cribsheet”. Emily is a Professor of Economics and Public Policy at Brown University.
P. Richard Hahn also joins the show as a guest interviewer. He is an associate professor of Statistics at ASU. His specific research interests include regression tree methods, causal inference from observational data, and foundations of statistics.
- Emily OsterProfessor of Economics at Brown University
Welcome to the Policy McCombs podcast, a data driven conversation on the economic
issues of today in this series. We invite guests into our studio to provide a highlight
of their work presented during a visit to the University of Texas at Austin Policy. Emma
Combs is produced by the Center for Enterprise and Policy Analytics at the McCombs School of Business.
We have two guests with us today. First, professors at the six, Richard Horne from Arizona State, Richard Worksman,
causality in decision making. And he joins me today to help with this interview. Now, turning to our main guest,
our guest today is Emily OSTER, professor of economics and public policy at Brown University. Emily
joined us to talk about her two bestseller books, Expecting Better Why the Conventional Pregnancy Wisdom is Wrong
and What We Really Need to Know. And the More recent creep sheet, a data driven guide to better and more relaxing parenting
from birth to preschool. Emily, welcome to Policy McCombs. Thank you for having me. So let’s start with the story
behind the writing of this book. How do you tell us how an economist gets gets to display? So writing
this this two books. Yeah. So so this short answer is that I got pregnant,
that I was an economist, and then I and then I was a pregnant, an economist. And I got really
into kind of what I can only describe as using my job in my pregnancy.
And so there were a lot of things that came up when I was first pregnant. Why I wanted to understand
what is the evidence behind this restriction or this recommendation or this rule.
And I was spending a huge amount of time at home basically doing doing things that were sort
of like what I would do at my jobs or reading academic papers, trying to think about decision making
and good decision making in the face of the data that we that we have. And
and that was me that was really in the service of my own pregnancy, not in the service of of writing anything like
this. But I had always kind of liked to write for a more general audience and think
about how do we explain the ideas that I think are really valuable about from economics, from data,
from statistics. How do we explain that to two people and how do we help them use
those data to make good decisions? And so at some point and this part, I
have to say, is a bit murky in my head at some point. I moved from doing this for my own pregnancy
to trying to actually write something for for a general audience. And what we found
was I really liked doing that. I really like the process of of writing in this way. And then I
sort of wrote some and then I had an agent and then I and then I had a book, got a
yadayada. And so then I published the first book, which is now is came out in 2013.
And I had had my daughter, who was born in 2011. And then
I was fairly sure I was I could write a second book because it’s a lot of work to write
a book and it’s had a lot of work to have a kid. And it seemed like maybe you doing both of them was not
was not really feasible. I think the other thing is that I you know, with your first kid,
I know you have two of them also. But I feel like with my with my first kid, every decision
about the kid was like made in this sort of totally like chaotic, haphazard way.
So we would just be like constantly like trying this, trying this and sort of obsessing about a lot
of really tiny decisions that not seem ultimately important in in retrospect, it was very hard to focus
on kind of what are the big things we’d want to make choices about. We had a second kid,
you know, with your second kid, at least for me, I was like a lot more relaxed. Not I’m not a relaxed
person, but relatively I was relatively more relaxed. And it was much easier
for us to focus on on kind of what are the main things I would want evidence about. And it was then much easier
to see. Okay, well, what kind of book what I write that would help people make choices about those those things.
And so then, you know, at some point a few years ago, I decided maybe it was time to time to
write another one. And so crib sheet is the kind of more or less the sequel sort of looking at the same
kinds of what does the evidence say? How do you make good decisions around early, early
parenting? So did this combination of data and economics. You talk a lot, a lot in the book
and trying to separate associations from causality. So give us a couple examples
that comes to mind when I’m thinking about crib sheet that I found very instructive in not only talking about the
evidence of this particular issue, but more to describe the endeavor overall to connect. Is
it for me that information to the general public is about breastfeeding and peanuts? That doesn’t mean those two
chapters. I mean, it’s a good summary of what what goes on there and is seeking causality. Yeah. So so let’s sort
of start with the question of breastfeeding. So when you sort of think about breastfeeding and the kinds of ways that breastfeeding
is talked about for four new parents, you really get a lot of the message that like breast milk, breastfeeding
is is the best. It’s good. The thing that is going to make your kid not only healthier and
better in the first year, but really forever is going to have these like very long term impacts impacting
their I.Q., their obesity, other kinds of other kinds of illness. And
this sort of basic issue with with data on breastfeeding is a lot
of the data that we have comes from studies which compare the outcomes for kids who are breastfed
to kids who are who are not in what we think of as a sort of correlation association
kind of way. So like to take a concrete example saying about something like IQ. So it is certainly
the case if you compare kids who are breastfed to kids who are not and you look at differences in. Their IQ, you will see differences
in IQ, some of them are very large, but when you look at the
characteristics of a parents, you will find that the moms who choose to breastfeed,
who breastfeed for longer, tend to be better educated. They tend to be richer, they tend to have more other
resources, they tend to be white. They have a bunch of other features which we know in the
data is also correlated with outcomes for their kids. So we know that maternal education
matters for kids test scores, and we also know that it matters for breastfeeding.
And so it’s hard to isolate the impacts of breastfeeding from the impacts of all of the other things that are
that are that are different. And so so in the book, I try to kind of go through and help
people understand how might we get better evidence on that. So we have millions and millions.
We have tens, maybe hundreds of studies of these correlations between
breastfeeding and and good and good outcomes for for kids. But we don’t have that many
studies which do a better job on causality, but we do have some. So, for example, there’s one
kind of randomized controlled trial of of breastfeeding, which has some of its own issues. But
but does does it bit better on this? And then we have actually a number of
of what we call sibling Fixx Effect studies. So studies where they actually compare to
kids within the same family and they look one of whom is breastfed and one of whom is not. And they look at
whether there are differences in IQ that show up there. So if you thought it was really the breast milk as opposed to other
characteristics of the mother, you would expect the differences to show up there. And they they don’t. So kids who are
when one siblings breastfed, one sibling is not, you really see basically no difference in their in their IQ. So
so I sort of tried to use that as a frame to help people think about, you know, what really are
the benefits of breastfeeding. And I think in the end, when you do that in a in a sort of,
you know, meta way for all of these outcomes, you do find that there are some things where breastfeeding does seem to benefit
the kids who particularly some early life digestion, maybe some early life allergy stuff
seems to be improved by breastfeeding. There actually seems like there’s maybe some evidence of reductions
in breast cancer for the mother, sort of long term, long term impacts. But on
a lot of these kind of long term impacts for kids like IQ and obesity and other kinds of health. The best data just does not
support those those things. So I think the message of sort of breast is best,
you know. Yeah, it’s it’s best, but isn’t it maybe not as as much best as
might be implied by the phrase breast is breast is best. So it’s kind of a classic
sort of like coat correlation versus causation, causation space.
So seasonally the peanuts are sort of. I agree it’s another interesting place. So. So here the question
is, is how you expose your kids to to allergens.
So peanut allergies have gone up a lot over time. There has been sort of historically and not by historically.
I mean, this is the advice I was given with my 8 year old. You were told we were told not to
expose our kids to peanuts, because if because that would make them more likely to be allergic or you wanted
to. You know, we didn’t. And so when they did, if they did have a bad reaction, they were older.
There was sort of like some stuff in that in that space. But at some
point, somebody did a very simple study where they just looked at the difference between
a peanut allergy rates between kids in Israel and kids in the UK. So Jewish kids in the U.K. and
kids and kids in Israel. And they found that basically rates of peanut allergies are way, way, way lower in
Israel than they were in the UK. So this is just like a correlational study and it’s fact. It’s like in some ways
even worse than the breastfeeding self because it’s like literally like a cross-country comparison. It’s like Israa. I mean,
could could you imagine something like, oh, I’m sure that that’s the only thing that’s different. You know, Israel versus the UK,
like a lot of Jewish kids in both places. Yeah. OK. And that really fixes you’re. Exactly. It fixes your problem.
But what they say in the in the in that paper, their theory is that Israeli kids eat
a lot of this peanut snack as like a 4 as an early foods is a peanut. Michael Bomba is very
popular and in Israel is like a Puft. My daughter loves those. Yeah. I mean, it’s delicious.
Who doesn’t like peanuts? And so kids kids are eating a lot of this at very young ages. And so they had this idea that, like, okay,
exposure is, you know, is good for preventing allergies. What’s
great about that literature is that they didn’t stop there. So if they sort of just had that, you’d be like, OK, what can we what
can we learn? But then they actually did a randomized controlled trial where they randomly exposed some kids
to peanuts early and some kids not. And the effects there are huge like the difference
in allergy development is like 70 percent.
And so basically exposing kids to peanuts. And I think it turns out other allergens as well
early on, like pretty much, you know, for months, kind of as soon as you’re exposing them to any foods
that turns out to be very good at preventing allergies. So that’s a kind of a good example
where we sort of started with the kind of evidence that you would say. Really? We would use
this as suggestive evidence and not as the as the sort of all piece of evidence. And then
they actually did the next thing, which is they like real randomized control trial where you can be more confident about causality.
So in the case of breastfeeding, that’s not they haven’t done that. The best we have is the sibling
study. So there is one randomized control trial of of breastfeeding. It was run
in Belarus in the 1990s. It’s it’s pretty big
and it’s an encouragement designed. So basically, they they encourage some women to breastfeed in some women.
They don’t encourage them and they get pretty large, a pretty large wheezy first stage. So pretty large
difference across groups in in their incentive of breastfeeding.
Both ever breastfeeding, but especially sort of continuation. So so people views, S.I.D.
You look at outcomes like both sort of short and long term outcomes. And there
you see very consistent with what you see from the sibling studies, you see sort of some evidence
of kind of digestive benefits, maybe some rash reduction early on,
but you don’t see sort of consistent differences in in IQ or test scores later. You don’t
see differences in obesity or other kind of health measures later. So on the one
hand, I think it’s a good it’s it’s a good study because it’s randomized.
It also isn’t necessarily the most comparable
setting to the sort of current kinds of questions people would ask. In the U.S.,
so, you know, it’s now actually pretty old. So I would like to see that people do
more of that. It doesn’t seem to hear you describe it. It doesn’t sound particularly
equivocal. In other words, at this point, would you say that the
data is in and the effect sizes are definitely smaller than the
media would suggest? Yeah, I think that’s right. I mean, look, I always like I always want
more data. And I think that, you know, we mostly the estimates we
have at the moment come with a fair amount of error. So, you know, could. Like, I think
that they they suggest that the effects are smaller than the kinds of things that are that are stated in the media.
Does that mean that they’re zero? You know, I think that’s that’s like outside the range of what we can do with our with our
current. Sure. So but yeah, I mean, I I think we have a lot of there are some places
in the book where I say like, you know, the problem is we just don’t have a lot of data on this. There’s a place where we do have a lot of
data. And so that raises the question sort of why or why. It’s
doesn’t seem clearcut to people. And we were talking earlier about opportunity cost reasoning.
Can you speak to the way you think about that in terms of specifically the breastfeeding example?
Yeah. So, I mean, I think one with one way to think about it is let’s imagine that breastfeeding was something that was
like super easy. And you could do like just in a just it would be trivial.
So it would be you know, it’s like vitamin D drops or something, something where you just kind of like
put a little thing on your kids for head every morning. Then probably we would say you should do it
because there do seem to be some some benefits. Right. So so I think in in that world
it’s you would see why you would want to get people to do it. And I think a
lot of the approach to breastfeeding has kind of taken on that frame of
like, well, everyone should do this. So let’s just like let’s really spend a lot of time talking about how great it is.
Maybe we overstate the benefits a little bit. You know, so we just really want people to get care to do this. And
I think what that misses this is where it argument is like. Actually, this is both very difficult
for a lot of people, like literally very hard to do. But also it it is
very costly in terms of your of your time. So one of the things he will say about breastfeeding is it’s it’s it’s
it’s free. Like, it’s such a great deal because it’s free. That’s that’s like the crazy
like, you know, I mean, I sort of think about the the kinds of time that people spend, you know,
pumping nursing like time out of a labor force. Like, you know, even if you just
just think about the pumping time, like how many people realize if you want to exclusively breastfeed a baby and you’re at your job
like you need like 30 minute pumping breaks every, you know, three hours
and it’s not that easy to work while you’re pumping. You can do it, but
it’s distracting. It’s distracting. So it is exactly to emphasize
the cost. There’s a lot of costs that are unnecessarily think about ahead of time. Yeah. And and, you know,
I think that the to me, the most impressive aspect of both of your books is this is fit make people
face that issue. The fact that these decisions are not made in a vacuum. They have costs, they have opportunity cost to have
they have tradeoffs that you have to to to think about. And when you do, I guess,
what what is the advantage to do it? Do you do any calculations on that when the breastfeeding? Do you have a sense
of, you know, the typical labor force costs associated with potentially breastfeeding?
That’ll be very nice. I don’t think we I think that would be interesting to see. And I don’t think we have a great sense of it. I mean, in
part it’s very complicated by the fact that actually, like the people who are doing the most of this, the people are most
likely to breastfeed are actually also people for the high value of time. Right. So I think they’re sort of like
the like the breastfeeding rates are much higher among particuarly continuation of breastfeeding or much higher among more
educated, you know, higher income women where you might have thought that that would sort of go the other go the other
direction. And I think there’s another piece of this which is sort of different from the average training costs, which is just
about support where, you know, actually it’s not like it’s physically difficult
to get to sort of get started doing this. So one of the things you see is that actually breastfeeding initiation rates
at the moment are very, very high. So very, very, very large share of of women in the U.S.
report trying to breastfeed or doing it for, you know, a day or two. But when you look at continuation,
even, you know, for a few weeks their way, they’re way lower. So it suggests that there’s a
sort of like pretty big drop off where people try to do this and either just
decide it’s not for them or it doesn’t or it doesn’t work. And I think that that we spend a lot of time
telling people that they should do this and maybe not as much time like helping them do
it, which is another cost there, too. People get really frustrated and they feel that they are failing their kid and depressed.
Exactly. So you have had the use of the story of
your book, starting with your pregnancy, essentially that your interest research wise has been related to
the medical field for longer than that. Yeah, I’m sort of. Did you what was your reasoning?
Was this something you had always been interested in? How did that come about? Yeah. So I. I
when I was in when I entered college, I thought that I would be a doctor or like I’ve always been very
interested in research. And so I guess I thought it would be like a medical research answer some specific ideas about being a more
of a hard scientist kind of person. And then I am the summer after my freshman year, I had sort of two
jobs. I was like a like a kind of Part-Time research assistant for an economist, Chris Avery,
who’s the Harvard Kennedy School and was doing some stuff about schooling. And then I worked at a fruit fly lab,
which was like my you know, like I’m going to be like a like a scientist. And it was
awful. I mean, I I like I. There’s nothing wrong with this person’s fruit fly lab. And like,
obviously, like, we’ve learned a tremendous amount of from fruit flies. But what I learned is that I was not cut out
for this for this experience of kind of hard science, for a bunch of different for
a bunch of different reasons. So I sort of pivoted into and I pivoted and in more into into
economics. But I I’ve retained a lot of interest in in medicine. And I’ve sort of a lot
of if you look at a lot of my work, it has this sort of flavor of kind of
overlapping a little bit with with medicine and and kind of being
in reading a lot in the medical literature. So there’s. So I think part of what made it possible to
to do these books and to do the kind of research that led into them was that I had a pretty
good sense of the medical literature and how it like how one would sort through that
coming into into doing this, which is going to be true of everybody in economics. And what
has been your experience as a social scientist, a very quantitative one working
in a medical field? Has the reception been mainly smooth and positive
or, you know, so I mean, I think there’s kind of two
two sets of people that that I interact with a lot. One is like
doctors. And, you know, I would say have I have mixed but mostly pretty positive
relationships with people actually practice medicine. And it’s particularly true as the as
the books have kind of age. So I think I get more pushback on the first book when it first came out
because could talk about. But some of the stuff about alcohol and some of these other things. And I think the first book also feels
a bit more confrontational with sort of my own experience with the medical system during pregnancy
was very frustrating. And I think that comes out in the book a little bit. And that was like last is
much less true with with the second book. And so my kind of my all of my pediatrics experience has
been great. I love all the pediatricians that I encounter. And so so there
I I you know, I think I’ve had sort of more positive, positive interactions. I will say
when I I spend a lot of time in the epidemiology literature and this isn’t a comment
about interacting with epidemiologists, but or, you know, I would say sort of public health literature in
general. I find that literature incredibly frustrating. And I think that I just like
the the lack of really
taking seriously concerns about causality in analysis of observational data
is just. I like something I say. I don’t. Still
totally get it here. I know. I just I mean, I find it like
on the one hand, they’ve got a lot of RTT is like, that’s great. But then when they’re not using r-s.d, you know, we’re we’re
regressing them again. This is like I’m overgeneralizing. There is there are, of course, people in this space that take causality
seriously. But there are so much of this. You know, The New York Times is covering, you know, eggs kill
you, eggs don’t kill you. You know, eggs are great for you. Coffey’s the death coffee is going
to make you live forever. It’s like, you know, it’s it’s election. It’s all just selection. Different
kinds of selection. Yeah. So do you think that you’ll be. I mean, right now you’re
sort of a the representative of a certain type of
research in this space. And I was asking people because they knew I was going to talk to you. What should I ask
her? And one of the questions was, do you think that there are gonna be more people that do this type of work or you just always
going to be a. I hope so.
I mean, I would really like to see more of a push
in this towards like the credibility revolution in this in this
like medical literature space. So if you sort of think about the path of
of economics and I understand I’m not reveal here that like I’m an economist and I think that what we do is like the right stuff.
But that’s okay. That’s part of the thing. You know, I think that that over time we have
really improved our ability to do inference out of observational data and the sort of seriousness with which
people think about research design as a way to understand causality and observational data. I
would like to see more of that in these other spaces. And I think that, you know,
as people have started talking more about publication bias in P hacking and I
feel like that’s that’s in this, guys. You know, I look like GPA any he’s like there’s like a
I feels like there’s a moment where maybe we could push a little bit more on on
that. And I also feel like there’s more public interest in kind of thinking seriously
about about causality. You know, I feel like when I when reporters call me now
about to comment on medical studies, they are more skeptical
sometimes like they are sometimes we’ll call and be like, you know, but that’s election again right there. Are you
writing for. Yes, I agree. Yeah. Thank you. Thank you. Gala’s I appreciate that.
But it’s a question that I think Richard had as well. But but let me get there.
Is it. I agree with you that that that in the social sciences, economies are far ahead in terms of thinking hard
about causality, in observation studies and being very you know, they lead the way to other fields to
follow and has been happening for a long time. And here’s another field where it where it can take advantage of that. But
the framework of of a causal inference based on on no testing, for example, was not particularly
well suited for decision making. And that’s something that you bring in the book, this connection
between evidence, decision making, which is fantastic. That’s how we know in particular statisticians like Richard. I
like to think about the world as everything is decision making problem. So have you ever given any thought
to that issue and thinking about maybe even frameworks of inference that might actually be useful
to make the connection between the evidence we collect and the decision making process? No, it’s super understanding.
I haven’t thought that much about that. But I think you’re exactly right that sort of you could frame the whole the
you could frame the sort of the whole thing as kind of ultimately the goal is the goal is decision
making. The goal is not to know is there a causal impact of X on why it is like to know
like should I do behavior? You know, what should I basically show I do this behavior and
that and that can do. Yeah, that’s very no. The answer is I haven’t thought that much about that, but that seems
great. That seems like more like the set sort of statistician approach. Yeah. We’ve been we’ve been involved in
a couple of studies in psychology now where where, you know, it’s obvious that the traditional that the
credibility revolution going on there, there’s this huge endeavor and trying to make sure that we pre-register everything
that you know and almost like a completely overcorrection to the situations where
all the studies were not being able to be replicate and so on. But we tried to work on
with tools that hopefully would help people in thinking about, okay, what I learned from this r_s_t_
if I were to now run the second one, you know how reprioritized, for example, the groups that might be more susceptible
to the treatment and so want to think about it, but that’s a long endeavor is not it’s not an easy
thing. And I think the psychology of been open more open to receiving
that kind of tools that we been working with then quite honestly, the economists and
that at making a connection, it’s something that would be kind of neat. Yeah. I mean, there’s also a space for sort of more
like kind of a Bayesian approach to some of these things like
you what’s right? This is sort of like. Yeah. I mean, I think we we have a lot of in a lot of these spaces.
One of the things that seems kind of weird is like we’re kind of everything is sort of sort of some frequentist
thing. And so you’ll have like, you know, a. I was in pieces of like a huge amount of like biological and other
kind of Prior’s that basically acts does not affect wire, couldn’t you know. And then it’ll be like, well, at least one study where
X effects is effects. It’s like, well, you know, like, yeah, okay. But like, could you do you know. You know, real eBays.
And so I think that’s. And that feels to me like pretty important for the decision making piece piece of this year. Like
how do we bring these pieces of evidence together to oversimplify a bit. You must have noticed this when writing
that the old saw that absence of evidence is not the evidence of absence. And
you have to write around it all the time because you would find in your writing you would say it’s
statistically significant. We can’t show that it’s not zero. Right. But then in the very next sentence, you
have to basically say, but to the extent that we have estimated it, we think that it might be positive.
Right. And then you have to take them to account. It’s a difficult. The whole framework is difficult to talk about. I think we’ve all taught statistics
and it’s just a nightmare. Yeah, I agree.
I have some more questions. Just so. So
I think the two themes that are coming up here and I think it’s a correct me if I’m wrong, but causality
and then utility, there are the two things that come up. And,
you know, both of those things are hard. The utility one is kind of interesting because
there’s some evidence that people don’t actually know their utilities that well.
Do you think that like how in some of these problems, in some cases, when it has
to do with an infant death or something horrible like that, the utility function is kind of obvious. But I
think there’s a lot of cases, especially in your second book, that are sort of lifestyle things that’s really hard to
hard to say. Do you have an approach for that? What do you tell people when they ask you? I don’t actually. You know
my mind. Yeah. I mean, I think I I I think that a lot of progress
on this can be made by just being a little bit more deliberate about the weight,
about the fact that things are our choices. So,
you know, rather like I think you’re right that like when we think when we talk about utility as as economists or statisticians,
we had this sort of like you have some Udalls and you’re right. Like nobody knows what they’re what they’re they’re Udalls.
They’re you know, nobody is aware of their of their to know what that what that is. It’s like a Dr. Seuss.
Exactly. They read the star Udalls and the nonstory. But
but I do think sort of framing. Just just trying to sort of say, okay,
these are your two like these are your two choices and think about your life with this and with, you
know, without this or, you know, think about this in the context of other choices you
and you will make or what is the alternative? I think those are, you know,
just like there’s not a lot of deliberateness almost. I would say people are not deliberate about these
these choices. And I think sometimes that reveals sometimes that kind of
frame can can help reveal. I can’t reveal what people or
people want. Yeah. But, you know, there are a lot of there are a lot of things like, you know, potty training is like the biggest
thing in the book. Rush is like whatevs. You know, sort of like people like what is the right age to potty train? Basically,
like there is no right age. If you wait longer, it will take less time, but you will have to change woobie
diapers for longer. And, you know, if you try to do it earlier, like it’s going to take more time, they’re going to be on the floor
some of the time. But, you know, they then you’ll be then you’ll be done. And, you know, you need to like
interrogate what is your feeling on diapers and that’s it. But there’s no, like, Zellick
secret way to make that choice. So, I mean, you advocate a lot for for thinking
about collecting the data, thinking about the costs and then making a decision and for some of the
problems that you’re using in your books. You let off with do it, eat as a family
or do we eat separately and we take out and you know, it seems to me some of those cases
are distinct or different from the medical cases in that you get feedback immediately.
And so in those cases, is it profitable to encourage people to experiment? And you might say luck with
breastfeeding might to some extent be the same way to the extent that that you can less so. And
it seems to be a spectrum of experimentation versus we have to make this decision figuring out there you tell it and then
figure it out. Yeah. Yeah. I think that that sort of good decision making, if you sort of think about like how do we
atabaki people make decisions and you know, outside of the home context, good decision making would involve
thinking about the choice, making the choice, reviewing the choice, you know, having like a structured process
for making some of these these decisions. I think that sort of saying like you should experiment is maybe.
Exactly, exactly right. There are many things where it’s it’s a little like a little bit hard to
really like kind of experiment. And I think the other thing that happens is, is people
don’t think carefully about the sort of review piece of this. You just sort
of fall into like doing something. And then that’s the thing you’re you’re doing. And there isn’t a moment where you step back
and you’re like, okay, actually, is this like, is this something it’s working. You just wait until sort of things
crash around, down, around you and then you’re like, oh, god, that’s working like I thought
it would. Yeah. Just to switch, switching a little
bit to two other areas in which this approach of data and economics you think might be helpful. You mentioned a little
bit about the fact that that you would like to see more in the medical field. Did
you. Did you start a trend? I mean, do you think that there are other folks out there trying to bring that kind of approach to
areas that you didn’t expect or not? Anyways, I’m asking, have you read something recently, any interesting book
that sort of mimics that approach and try to apply to different areas? I mean, I think, you know, I’ve
I think that there’s you know, there are some people who have worked on kind of like school.
I mean, economist have gotten more into like writing popular books. Since I wrote this. I don’t think because
of me. But just like whatever. There’s been more of this or maybe a short me. It’s probably it’s I actually that’s
not telling me how to present. Nothing to do with tickety. Who goes by that guy.
No. But, you know, I think that there’s there are a bunch of spaces where economists are working where you can imagine sort of
translating that work in a way that’s sort of more direct into like it’s sort of like work, unschooling
or, you know, the. There actually is another economist parenting book by like The TimeStep
Key and Sloboda. I think that’s like about like kind of the theory
of of of parenting. So, you know, I think that there’s some there’s some
push in in that in that direction. But I don’t know that there’s anything quite in
this quite in this space. Yeah. OK. So.
So this is maybe the most confrontational question that I have for you, which is that
not everybody’s credit equal in their decision making skills. Sure. And, you know, this is certainly something
when medical bodies or institutions have to come up with policies. What
do they think? They kind of have to go for a lowest common denominator approach.
And so, in short, maybe more information is not always
better for outcomes. Surely you’ve thought about. Yes. I’m interested
to hear your thoughts. Yeah. So, I mean, I think I think a lot about this when we’re we’re kind of talking
about something like sort of safe sleep guidelines or even something, you know, like alcohol
and pregnancy, kind of like if we if we think if we agree, let’s us take safe
sleep. So there’s an issue of like, you know, should your kids their sleep guidelines urge kids you sleep like alone in their crib, not
in your bed with no stuff around. And in some pieces of those safe sleep guides are easier
to to do than others. It’s easy to not have bumpers in your crib. And I have a lot of pillows in there,
but it may be much harder to get your kid to sleep alone. And I think from the standpoint
of the of the of the AP, there’s kind of a feeling like, okay, we’re just going to we’re gonna
say these things with this is the safest way. And if everybody did this, that would be like kind of the safest
the safest thing to do. But what I think is is hard
about that messaging is that then if people can’t do that
or or don’t do it, you haven’t given them any guidance about how bad
it is to deviate in different directions. And so there’s a sort of piece of like,
yes, it would be best if everybody did this one thing. So let’s just tell them to do that one thing and not give them any information about
the other, the other things. But then if they choose not to be
in the box that you have given them, they they may deviate in a way that’s that’s really
bad. So let’s take the example of of a sleep’s. You tell people you can’t have your kids leaving your. It’s dangerous
to have your kids sleeping in your bed. But the truth about the data is that it
is probably slightly riskier to have your kids sleep in your bed. But that risk is really different
if you are sleeping like you know, if the adults are not smoking and not drinking.
If there aren’t a lot of other covers in the bed, if you sort of think if you’re.
So those are kind of the big factors. And what we don’t really tell people
that, we just say it’s not safe to do this. So then you can you can people sort of exit
the almost like accidentally do it or they’ll they’ll do it, but they they won’t do
it in the safest way possible because no one’s told them that there is a safer and less safe way. It’s just like, don’t do this while once I’m doing
it, I might as well like sleep on the sofa. But actually, no, sleeping on the sofa with your infant is like,
really, really, really dangerous. Way more dangerous than sleeping in a sort of the safest kind of co-sleeping
environment. So I think that’s the piece where we want to be a little bit careful about
this idea that just telling people the safest thing is right because it may be
very hard for them to actually achieve that. It’s notable that what you describe still
entails some processing of the data. It’s not like you’re just providing the tables. You’re actually giving a rank ordered
menu and you’re saying that one might be on the top. But then, you know, and it reminds
me of some colleagues of a one time colleagues of ours,
Dick Thaler, Cass Sunstein, their book Nudge, which basically
sounds sort of a similar process. Did you ever talk to them about your book in the context
of their idea of. Well, not no. I mean, I’ve talked to Dick a lot about this, a Richard a lot about the
about this book, about the books. But I haven’t talked to him on that. But I think you’re right, it sort of has this that kind of feel
good, some way to behavior take on all day. Is that. Well, no, we lay people utility functions make the decisions
because they have all these problems in their psychology. Therefore, we need to nudge them in some way. A be paternalistic to some extent.
Right. And I think your presentation in more traditional econ sense of like now mean here’s the data.
You have your utility, go ahead and empower yourself to make a decision, right? Yeah. I think I mean, I think in these species,
in both of the parenting, all the press, even more so in the in the pregnancy space, I think there is too
little credit given to people about their ability to sort of
like ascertain their utility functions, make decisions, think about risks and and benefits from themselves. And I
think that’s you know, I think that can border on disrespectful, honestly,
particularly in in those sort of pregnancy games that they’re also sort of like, oh, don’t worry, I’ll tell
you what to do. You know, you’re just like a little pregnant lady. You know, as opposed to to recognizing
that a lot of women are actually perfectly capable of making choices for themselves,
has to be has to be sort of somewhere in between. Right. Because if it’s in respect versus exposed
outcomes, you know, I mean, is I really like the articulation of having a menu like that. Seems
like I’d never thought about that in particular. That seems like a. MiddleGround. I think there’s a way.
I mean, there’s also a lot of these things like there’s a way to like be be respectful of people’s decision
making and timing while not just being like do whatever, you know, sort of like, you know, Italian,
basically. Like, I recommend that you you don’t make this choice about your birth. And, you know,
here’s why I think that’s important. Is that is it something which, of course,
doctors should be empowered to say, in fact, is like their their job, but maybe different than saying
like, you have to do this or your kid is going to, you know, have some terrible,
terrible outcome. I mean, I wouldn’t. When my son I’ll tell you, when my my son was born,
he had jaundice and they came in like me, like a couple of days, you know, we went home.
Somebody came. They took his blood. They called the doctor called us. And she was not a regular doctor.
And she was just basically like, you know, your kid is this is his number.
And I said, okay, can you help me understand, like, you know, like what? What exactly? Like,
how do you where’s the data on the cutoff? My first kid hadn’t had this in have any data like where’s the data on the cutoff? Like,
how do I understand? How do I understand this? And she was just like, you know, I’m telling you what to do if you don’t bring
your kid in to back into the hospital like he’s gonna have brain damage. And, you know, that was there was.
And of course, that was like just to be clear in our case, 100 percent wrong. Like there was nothing.
I mean, jaundice can cause brain damage, but we were like distantly far from any number
that would even actually would even recommend hospitalization, let alone be a risk for.
But it was sort of like presented in this way. That was that was really
not respectful of the fact that like perfectly reasonable people could differ on this. And then, in fact, the question
of just, can you help me understand where these cut offs come from? And you know, how you
process this like that, that this sort of the frame was kind of like, I can’t believe you would even suggest
that you would have an opinion about this and that you would even think to ask this question. I told you what to
do. I’m the doctor. There has to be infuriating. Yeah, it was. And it was
really I was so angry at actually the next day this person sort of what I ETSA
like I did bring him back to the hospital because of course, you know, three days postpartum and somebody tells
you if you don’t bring your kid by glass, he’s got brain damage. I guess you’re breaking my glass bottle. That’s the vulnerability. Also, that’s
really irresponsible. Was like, what are you going to. You know, what are you in it? So we brought it back to hospital. And, you know, this is he was finding
really have any edge on us. And but then the next day, she did come in and say and she. She
had like figured out. I mean, she came in the next morning before rounds and told
me, oh, I figured out who you were at, how. Now I figured out who you are. And I think we can work together.
And I just want to be clear. This is not like a person who is a pediatric. This is
like some hospital. You know, this was like some random person, but it was a very it was sort of very telling mom
around to be like, okay. Like, I understand that. I I know why you’re
reacting that way. But just to be clear, it doesn’t matter whether I’m a professor or a parenting book
offer. Anything else like that is a perfectly reasonable question. What cut off did you enter
and where? Like, where are you getting your your evidence from? That’s a question that could be asked by anybody.
You should be respectful to people, even if they’re not going to go on the radio and the five guys to talk
about your behavior. Time for one more. Yeah. There’s a few
more. Go ahead. So here’s a half baked idea of mine. One of the things that
I’m interested in my own research is heterogeneous treatment effects. So not every drug affects everybody
else. Does that translate to. Yeah. Yeah. Well, it’s just different. People have different responses to
different treatments. And one of the things that’s occurred to me is that almost all of the medical
literature has been focused on average treatment effects. There’s good reasons for that and not so good reasons for that.
But in particular, it as a thought experiment, I was wondering maybe
getting anecdotal advice with multiple anecdotes. So from people close to you
who are more like you can sometimes be better than knowing about an underpowered
medical study based on a population that might not be like you and
as a as a logical matter of course that’s the case. But I
don’t know to what extent it’s real. So if you just what are your thoughts on heterogeneous effects and whether or not it
salvages folk wisdom in any circumstances? So this this is a sort of source
of a lot of when you ask doctors like why do they rely so much on their own experience in giving
in giving advice? You know, either sometimes the answer is just like mine. That’s how I that’s how I do
it. But I think if you if you kind of there are many thoughtful people who would say, look, the reason I give this
advice is because even though I know these the sort of average treatment effect from these studies,
I have a lot of expertise on my population and I am kind of triangulating between the sort of
like trying to to use my population to figure out the treatment effects that are specific
to the people that I work with and kind of combining that in some way with these others.
So, I mean. I think the zouk, super important, super important space, it comes off all the
time, even in in the sort of RTT, a good RTT evidence around something like obstetrics. So you’ll have like a
good r-s.d that is about the impacts of induction on C-sections
or something. This is like the most recent thing that that has come out. But if you sort of dig into like that’s
a those are can be great studies. They’re big, they’re powered that, you know, they’re they give you a lot of evidence.
They come out of fancy teaching hospitals. Right. So this sort of really big trial that recently
came out about induction C-sections, the the C-section rate in the trial, in the hospitals
and the trials like 18, 20 percent for the for the drug for the control group.
That’s, you know, the C-section rate in the U.S. is like 35 percent. So if these places are very selective
in a particular way, are those adverse treatment effects relevant for the for the other? I mean, is it a little
different than heterogeneous, but not really. I mean, it sort of says like your Yahtzee dieselgate for an average treatment effect for
the people who are willing to be in, you know, where the hospitals can do a protocol on this. In this
r-s.d. So, you know, I think the the idea of you
even using some of these tools that people have taken to try to kind of combine observational
like sort of battled like bias, observational evidence, but where you can get heterogeneity with
good, you know, average treat with with good randomised evidence where you can like that, that feels like
something where economists, again, have sort of probably also statisticians have like, you know, try to make some progress
and where where you could sort of do more do more here. Yeah. Okay. So that’s an
interesting I’m not really listening. I’m not sure I would have put like listening to your friends in this space.
Well, I see grandma, as you know, was there in the sense that, you know, we’ve seen this here
in this group. In some ways that somehow is different because I was imagining in particular a population that
is distinct in some very particular way, indigenous group, in a particular geography,
that the advice passed down from generation there might be contra the controlled studies, but might make
perfect sense in the context of that climate or that seems that seems totally.
That’s my one more. One more. Yes. Well, this is this is too easy. That
makes it go in the close. So medical info is gonna change. Have you signed yourself up for
a lifetime of doing edits day? So I’ve seen the book. So
we have updated the first book a few times and I think we’ll do like a much bigger update in the next couple
of years. And so, you know, I think I don’t know if I ever signed myself up for like
a lifetime of it, but I think as long as people are buying the books, I would like them to be.
I would like them to be updated. And, you know, I don’t know. It’s like it’s it’s fun.
It’s sort of fun to stay up on this. This literature. Well, for what was worth.
Thank you so much for writing this book. As a parent and as a husband, there were incredibly useful
for it for for our family. And I think a lot of families can take advantage of that. And thank you, Richard, for joining us today.
Thank you. Absolutely. Thank you, guys, for. Before we wrap
up, you can get more information in our medium page. Thanks for listening to Policy McCombs.
See you next time.