Latest / Wellness, Actually with Emily Oster & Perry Wilson, MD / Random Acts of Data with Dr. Anupam Jena

Transcript

Random Acts of Data with Dr. Anupam Jena · Wellness, Actually with Emily Oster & Perry Wilson, MD

• 0:05
  Speaker 1: Welcome to parent Data by Emily Oster. I'm Emily. I'm
• 0:09
  Speaker 1: a professor of economics at Brown University and a mom
• 0:12
  Speaker 1: of two. I write a newsletter called parent Data, and
• 0:15
  Speaker 1: on this podcast, sometimes I'll read it out loud, and
• 0:17
  Speaker 1: other times, like today, I'll interview fellow data driven parenting experts.
• 0:23
  Speaker 1: If you want more of my writing on parenting and pregnancy,
• 0:25
  Speaker 1: you can sign up for free at www dot parentdata
• 0:29
  Speaker 1: dot org. You'll get reader stories, Q and as, information
• 0:33
  Speaker 1: on our revamped podcast that we're rolling out in the fall,
• 0:37
  Speaker 1: and lots of other features delivered straight to your inbox. Today,
• 0:41
  Speaker 1: I'm talking to doctor Bachwojenna, host of the podcast Freakonomics, MD,
• 0:46
  Speaker 1: about randomness and health, when your kids should get a
• 0:48
  Speaker 1: flow shot, and how much of the parenting you plan
• 0:51
  Speaker 1: for might actually be based on chance. Here's Bachwujenna. Thank
• 1:00
  Speaker 1: you so much for.
• 1:01
  Speaker 2: Joining me, Thank you for having me.
• 1:03
  Speaker 1: So we're here today to talk about your new book,
• 1:05
  Speaker 1: Random Acts of Medicine. And I was thinking about how
• 1:08
  Speaker 1: to introduce you, and the thing that occurred to me
• 1:11
  Speaker 1: is that I think you are, in some ways the
• 1:13
  Speaker 1: fellow academic who I am most similar to in terms
• 1:17
  Speaker 1: of how I think like we work in these similar
• 1:19
  Speaker 1: spaces something between health and economics, and I think we
• 1:23
  Speaker 1: both basically agree that in the end, you can learn
• 1:27
  Speaker 1: everything from data, and that if you just look at
• 1:30
  Speaker 1: it right and you find the right data, like there's
• 1:32
  Speaker 1: some kind of secret answer there. But there are some
• 1:35
  Speaker 1: ways in which we're different, and the main one is
• 1:37
  Speaker 1: that you are an actual medical doctor in addition to
• 1:40
  Speaker 1: being trained as an economist. And my sense is that
• 1:45
  Speaker 1: we are both pretty critical of the data approaches that
• 1:48
  Speaker 1: are used in the medical literature, but that you are
• 1:51
  Speaker 1: critical from the inside and I'm critical from the outside.
• 1:56
  Speaker 1: And in a sense, I think the book is really
• 1:57
  Speaker 1: a presentation of how data could be used better in
• 2:01
  Speaker 1: those questions, rather than just being a criticism of how
• 2:05
  Speaker 1: people use data. So what I'm hoping will do today
• 2:08
  Speaker 1: is actually talk about some of the examples from the book,
• 2:10
  Speaker 1: but then also sort of step back and talk a
• 2:12
  Speaker 1: little bit about these issues of methods and how we
• 2:15
  Speaker 1: might improve the methods that are used to estimate causal
• 2:19
  Speaker 1: relationships in medical data. So before we do that, can
• 2:23
  Speaker 1: you just please introduce yourself.
• 2:26
  Speaker 2: Well, first of all, I thought you were just going
• 2:28
  Speaker 2: to introduce me as a Sunjay group of health policy.
• 2:30
  Speaker 2: But you gave me a longer introduction than that.
• 2:32
  Speaker 1: But I don't even think of you as a Sunjay group.
• 2:36
  Speaker 1: I think of your Is that how you introduce yourself?
• 2:38
  Speaker 2: Well, no, it's not. But I'm seeing myself and my
• 2:41
  Speaker 2: video here talking to you. I'm like, wait, if I
• 2:43
  Speaker 2: just got my eyebrows, eyebrows.
• 2:46
  Speaker 1: Are more significant?
• 2:48
  Speaker 2: Yeah, of course significant. Well, my name is Babu Jenna.
• 2:51
  Speaker 2: I'm an economist and a physician at Harper Medical School,
• 2:55
  Speaker 2: and I see patients at Massachusettsenteral Hospital. That's who I am.
• 3:00
  Speaker 1: So this is a parenting podcast. Do you do you
• 3:02
  Speaker 1: want to tell us? Do you have like children?
• 3:04
  Speaker 2: Oh? I do? Yeah, I have two. I have a
• 3:06
  Speaker 2: five year old, uh and an eight year old and
• 3:09
  Speaker 2: uh yeah, they're They're a joy. You know, I'll just
• 3:13
  Speaker 2: say this, I know this is yeah, you talk to
• 3:16
  Speaker 2: a lot of parents. Uh. The one thing I think
• 3:19
  Speaker 2: differently about now that ever before is I remember when
• 3:22
  Speaker 2: I was growing up and we didn't have cell phones
• 3:24
  Speaker 2: back then. Anytime I go to a friend's house, my
• 3:27
  Speaker 2: parents would make me call them from that friend's house
• 3:31
  Speaker 2: when I got there. And you know, if I was
• 3:32
  Speaker 2: seeing you know, a potential girlfriend of that kind of thing.
• 3:35
  Speaker 2: That's so embarrassing, right, I say, oh, I've got to
• 3:37
  Speaker 2: make this important phone call, and I just whispered to
• 3:39
  Speaker 2: my parents. Now I know what they're going through, because like,
• 3:42
  Speaker 2: if my daughter doesn't come back five minutes when you know,
• 3:45
  Speaker 2: five minutes ago, then she's supposed to. I'm getting nervous.
• 3:47
  Speaker 2: So it's come full circles.
• 3:50
  Speaker 1: So now you're going to make your kids call you.
• 3:52
  Speaker 2: Well, now now we've got like the uh.
• 3:54
  Speaker 1: Now you're going to track them on their phone exactly
• 3:56
  Speaker 1: like that ship in there. You put a ship in
• 3:58
  Speaker 1: their shoe exactly. No, okay, all right, So let's start
• 4:04
  Speaker 1: with you giving a very broad overview of describing your work.
• 4:08
  Speaker 1: When you look for problems to work on, what are they?
• 4:12
  Speaker 1: What are you looking for? What is this research about?
• 4:15
  Speaker 1: Because the book is really an overview of your research.
• 4:17
  Speaker 2: I would say that's right, Yeah, it's an overview of
• 4:20
  Speaker 2: my work. In recent years, I've had the pleasure of
• 4:23
  Speaker 2: working with someone named Chris Worsham who wrote the book
• 4:25
  Speaker 2: with me. He's a critical care doctor at MGH. But
• 4:28
  Speaker 2: the book is about how chance occurrences affect our health,
• 4:31
  Speaker 2: and it could be anything from you know, living in
• 4:35
  Speaker 2: a city that's hosting a marathon and you not being
• 4:38
  Speaker 2: able to get to the hospital in time because the
• 4:39
  Speaker 2: roads are closed. It's a totally random thing. It affects
• 4:42
  Speaker 2: your health, and we show how it does. But it
• 4:44
  Speaker 2: also teaches us something about how healthcare works, and I
• 4:47
  Speaker 2: can talk more about that my research in general, I'd
• 4:50
  Speaker 2: say I use big data, I use the tools of economics,
• 4:52
  Speaker 2: but I'm really interested in questions that are sort of
• 4:55
  Speaker 2: like freakonomics meets medicine. That's the sort of mo So
• 4:59
  Speaker 2: I want to.
• 4:59
  Speaker 1: Focus on one chapter of the book, which is the
• 5:02
  Speaker 1: one about kids and flu, partly because we do parenting
• 5:06
  Speaker 1: and partly because actually I think that was the one
• 5:08
  Speaker 1: chapter of the book where I didn't know about this
• 5:11
  Speaker 1: research before I before I read. I read the book,
• 5:14
  Speaker 1: and the chapter starts, like many of the chapters, with
• 5:18
  Speaker 1: effectively an observation from your own experience. And so when
• 5:22
  Speaker 1: you say it's random acts of medicine, sometimes it feels
• 5:24
  Speaker 1: like it's random acts of things that BAPU noticed and
• 5:27
  Speaker 1: then wanted to understand. Yes, yeah, can you talk about
• 5:31
  Speaker 1: what is the observation that motivated that chapter?
• 5:34
  Speaker 2: And by the way, we pitched that title. First, that
• 5:36
  Speaker 2: that didn't work at long. It's a little long. Yeah.
• 5:41
  Speaker 2: So the basic thing that happened to me, so are
• 5:43
  Speaker 2: we we have a five year old eight year old.
• 5:44
  Speaker 2: Our five year old has an August birthday, and so
• 5:47
  Speaker 2: like many young kids, we take him to the doctor
• 5:51
  Speaker 2: around his birthday. So he's born in August. So we
• 5:53
  Speaker 2: go to his three year old checkup and I'm walking
• 5:56
  Speaker 2: out of the office. It's like middle of August. And
• 6:00
  Speaker 2: as we're walking out, the nurse says to me, come
• 6:03
  Speaker 2: back in a few weeks to get the flu vaccine
• 6:07
  Speaker 2: for your son. And I was like, okay. And the
• 6:09
  Speaker 2: first thing that occurred to me as wow, had he
• 6:11
  Speaker 2: been born just two weeks later or three weeks later,
• 6:15
  Speaker 2: we would have gotten the flu shot in the office
• 6:17
  Speaker 2: that day, and we didn't. So that's the first part
• 6:22
  Speaker 2: of the story. The second part of the story is,
• 6:24
  Speaker 2: all right, well, you know, I'm a motivated parent. I
• 6:26
  Speaker 2: wanted to get him the flu shot. I probably spent
• 6:29
  Speaker 2: about three hours trying to get that thing done. The
• 6:32
  Speaker 2: pediatrician's office booked up pretty quickly, like within hours. I
• 6:36
  Speaker 2: think of when I try to call them or when
• 6:38
  Speaker 2: they opened up the slots. I was calling CVS, Walgreens.
• 6:41
  Speaker 2: These are places that back then would not offer flu
• 6:46
  Speaker 2: shot to a three year old because they're too young.
• 6:48
  Speaker 2: And so finally I was able to get him into
• 6:50
  Speaker 2: the pediatricians' office. But it made me think that something
• 6:53
  Speaker 2: as random as when he was born could have a
• 6:56
  Speaker 2: dramatic impact on the likelihood of him getting a vaccine.
• 7:00
  Speaker 2: That same observation generalized to all kids in this country.
• 7:06
  Speaker 1: And so here in the language of how you started
• 7:08
  Speaker 1: with your motivation, the randomness is in the when your
• 7:12
  Speaker 1: kid is born, and the question is to what I
• 7:15
  Speaker 1: sent us that influence whether they get a flu vaccine
• 7:18
  Speaker 1: and then presumably their ultimate health, the chance they have
• 7:21
  Speaker 1: a serious flu.
• 7:22
  Speaker 2: That's right.
• 7:24
  Speaker 1: Okay, So you have this idea, and you spend all
• 7:26
  Speaker 1: this time, and now you're annoyed because it's taking you
• 7:28
  Speaker 1: three hours to get your kid this vaccine, and so
• 7:31
  Speaker 1: you decide you're going to write a paper about it.
• 7:34
  Speaker 1: That's roughly correct.
• 7:35
  Speaker 2: That's I was annoyed with love, but just yeah, yeah,
• 7:38
  Speaker 2: everything goes correct. Yea, And I want I'm annoyed.
• 7:40
  Speaker 1: With your kid. You're annoyed with the PDA. Okay, no, no, no,
• 7:42
  Speaker 1: it's not his fault. So then where do you go
• 7:45
  Speaker 1: from there? I mean there's a sort of immediate approach
• 7:49
  Speaker 1: that one could say, which is, well, I'm going to
• 7:53
  Speaker 1: compare kids who you know, get the flu shot of
• 7:55
  Speaker 1: the pediatrician to not. I mean, there's a bunch of
• 7:57
  Speaker 1: ways you could approach this. What is the step? What
• 8:00
  Speaker 1: is the next step for you in terms of turning
• 8:03
  Speaker 1: that observation into something that would be of interest to
• 8:07
  Speaker 1: an academic audience.
• 8:09
  Speaker 2: I think the next step for me, and probably the
• 8:11
  Speaker 2: next step for people like us, is to think, what
• 8:14
  Speaker 2: is the experiment? The experiment is that a child is
• 8:17
  Speaker 2: born in August versus born in September. They're otherwise pretty similar,
• 8:21
  Speaker 2: and we can see that they're similar, but one group
• 8:24
  Speaker 2: of children with August birthdays is perhaps less likely to
• 8:27
  Speaker 2: get the flu shot. And so what we did is
• 8:29
  Speaker 2: we looked at very large data from insurance companies. Anytime
• 8:33
  Speaker 2: you or your kid goes to the doctor and the
• 8:36
  Speaker 2: doctor bills the insurance company for that visit, there's a
• 8:39
  Speaker 2: record of that, and that's because they get paid for
• 8:42
  Speaker 2: that visit. So we know, for example, when kids were
• 8:45
  Speaker 2: getting flu shots, and the nice thing here, or maybe not,
• 8:49
  Speaker 2: the nice thing is that these records, these vaccinations, they
• 8:52
  Speaker 2: show up in the insurance data because a three year
• 8:55
  Speaker 2: old could not go to CVS or Walgreens and pay
• 8:57
  Speaker 2: out of pocket or the parents couldn't pay out of pocket,
• 9:00
  Speaker 2: so there's an insurance record of that. And then all
• 9:02
  Speaker 2: we did is very simple. We just looked at kids'
• 9:05
  Speaker 2: birth months, and the insurance company knows that they know
• 9:08
  Speaker 2: when you're born. And we showed that kids who were
• 9:11
  Speaker 2: born in the summer are about fifteen percentage points less
• 9:16
  Speaker 2: likely to get a flu shot than kids who were
• 9:18
  Speaker 2: born and let's say September, October, November, December. And we
• 9:22
  Speaker 2: did a couple of other things just to sort of
• 9:24
  Speaker 2: show them mechanism. And one of the things that we
• 9:26
  Speaker 2: showed is that kids get their flu shot in the
• 9:30
  Speaker 2: pediatrician's office often around the time of that annual visit,
• 9:33
  Speaker 2: so that's sort of an important part of the mechanism.
• 9:35
  Speaker 2: But the main finding was that if you have a September, October,
• 9:39
  Speaker 2: November birthday, you're much more likely to get a flu
• 9:41
  Speaker 2: shot as a.
• 9:42
  Speaker 1: Kid, and then are there better outcomes for those kids?
• 9:45
  Speaker 2: Yeah, So that's the next question is like if the
• 9:47
  Speaker 2: flu shot works, then you would think that the kids
• 9:50
  Speaker 2: would be less likely to get the flu, and that
• 9:52
  Speaker 2: is true. We show that they are less likely to
• 9:54
  Speaker 2: get the flu. The next question may not be obvious
• 9:58
  Speaker 2: to people at least can see, but if I tell
• 10:00
  Speaker 2: you which, I will tell you it's then obvious. Which
• 10:03
  Speaker 2: is all right, Well, are the parents or the family
• 10:06
  Speaker 2: members of those kids differentially affected? And the answer is yes.
• 10:10
  Speaker 2: So if you have a child with a summer birthday,
• 10:12
  Speaker 2: they're less likely to get the flu shot, they're more
• 10:15
  Speaker 2: likely to get the flu, and then their family members
• 10:17
  Speaker 2: are also more likely to get the flu.
• 10:20
  Speaker 1: So what I think is important, I mean, the findings
• 10:23
  Speaker 1: are important, but I think it's worth stepping back to
• 10:26
  Speaker 1: the method that you're using here because to answer this
• 10:29
  Speaker 1: last question, so you could you might be interested. In fact,
• 10:33
  Speaker 1: many people would be interested in the question of to
• 10:35
  Speaker 1: what extent does a kid getting the flu shot affect
• 10:39
  Speaker 1: their parents getting the flu or affect older relatives getting
• 10:42
  Speaker 1: the flu. You know, that's a question of a lot
• 10:43
  Speaker 1: of policy interests. We could ask it about COVID, we
• 10:46
  Speaker 1: could ask it about the flu. A simple way to
• 10:49
  Speaker 1: do that analysis is to compare the flu getting the
• 10:53
  Speaker 1: flu status of older relatives of kids who do and
• 10:57
  Speaker 1: do not get the flu shot. So you could take
• 11:00
  Speaker 1: actual data on whether kids get the flu shot, compare
• 11:03
  Speaker 1: kids who have the flu shot and kids who don't
• 11:06
  Speaker 1: have the flu shot, and then look at their parents.
• 11:08
  Speaker 1: The problem with that, I mean you and I know this,
• 11:11
  Speaker 1: but just to be clear, the problem with that is
• 11:14
  Speaker 1: that on average, the kind of kids who get the
• 11:18
  Speaker 1: flu shot, the characteristics of kids whose parents get them
• 11:20
  Speaker 1: the flu shot, are different than those who do not.
• 11:24
  Speaker 1: That the choice to get your kid the flu shot
• 11:26
  Speaker 1: is connected to your parents, your income, your resources, parential education,
• 11:31
  Speaker 1: a lot of other things that also impact whether the
• 11:35
  Speaker 1: parents are going to get the flu independent of their kids.
• 11:38
  Speaker 1: What is very clever about your analysis is that it
• 11:41
  Speaker 1: avoids that problem by using these birthdays, right, you're exploiting
• 11:46
  Speaker 1: this randomness in which kids are born to what we'd
• 11:51
  Speaker 1: say technically to identify to see the impacts of the
• 11:55
  Speaker 1: flu shot on kids and then on their parents. And
• 11:59
  Speaker 1: that element of that is the direction for causality. That
• 12:04
  Speaker 1: is the thing that lets you say that this is
• 12:07
  Speaker 1: a causal relationship and not just a correlation.
• 12:11
  Speaker 2: That's exactly right. And let me add two more things,
• 12:13
  Speaker 2: and then I'll ask you a question. The first is
• 12:17
  Speaker 2: that it's a good way to show the causal effect
• 12:20
  Speaker 2: of getting the flu shot. Totally agree. And there's a
• 12:23
  Speaker 2: second finding, which is, well, you know, if you're interested
• 12:25
  Speaker 2: in how healthcare systems should be structured, you might be
• 12:28
  Speaker 2: interested in knowing that something is arbitrary as your birth
• 12:32
  Speaker 2: date has a pretty large effect on the likelihood that
• 12:35
  Speaker 2: children get flu shots, children then get influenza, and their
• 12:38
  Speaker 2: family members get influenza. So it speaks to how we
• 12:41
  Speaker 2: might think about redesigning the process by which we vaccinate kids,
• 12:46
  Speaker 2: not necessarily just tying it to the visit that they
• 12:49
  Speaker 2: happen to be able to get to because it's in
• 12:52
  Speaker 2: the fall, versus not happen to be able to get
• 12:55
  Speaker 2: to because it's in the summer. So I think those
• 12:59
  Speaker 2: are the two. Let me tell you the following though.
• 13:02
  Speaker 2: So we've got some other work which is not in
• 13:04
  Speaker 2: the book. It's not yet published, but it uses the
• 13:07
  Speaker 2: same approach to answer a different question. Can I tell
• 13:10
  Speaker 2: you about that? It's god quite related though. All right, So,
• 13:13
  Speaker 2: about a year and a half ago, I'm on a
• 13:15
  Speaker 2: zoom call with the rest of the faculty in my
• 13:19
  Speaker 2: department at Harvard and the chair of my departments asked
• 13:23
  Speaker 2: me what I've been up to, and I said, I
• 13:24
  Speaker 2: just know. My arm's a little bit sore. I just
• 13:26
  Speaker 2: got back from CVS and got the vaccine. And she
• 13:30
  Speaker 2: says to me, well, this is like maybe end of August.
• 13:33
  Speaker 2: And she says to me, well, Bapu, why did you
• 13:35
  Speaker 2: get the vaccine so early? And I said to her, well,
• 13:37
  Speaker 2: what do you mean? And she said, well, aren't you
• 13:39
  Speaker 2: worried about the immunity waning off or waning before let's say,
• 13:46
  Speaker 2: end of December January February. And I said to her, Barbara,
• 13:50
  Speaker 2: I didn't even think about that. I was there getting
• 13:52
  Speaker 2: some doritos, and the flu shots are available, and so
• 13:56
  Speaker 2: I said, I'm going to get my flu shot.
• 13:57
  Speaker 1: I didn't counteracts the doritos.
• 14:00
  Speaker 2: Correctly, or I would think them as more compliments. Actually, yeah, yeah, yeah.
• 14:07
  Speaker 2: So she asked this question, and I thought to myself,
• 14:09
  Speaker 2: all right, well, suppose I wanted to figure out the
• 14:12
  Speaker 2: optimal timing of the flu shots. Should I get it
• 14:15
  Speaker 2: in September, October, November, December to a give my body
• 14:22
  Speaker 2: the time to develop immunity against the influenza virus because
• 14:26
  Speaker 2: that takes time, but also to make sure that the
• 14:28
  Speaker 2: immunity doesn't go away because it's been too long since
• 14:32
  Speaker 2: when I was vaccinated and when I started getting exposed
• 14:35
  Speaker 2: to flu. And the question that you just posed a
• 14:38
  Speaker 2: moment ago is really germane here. If you look at
• 14:42
  Speaker 2: people who get vaccinated earlier in the season and look
• 14:46
  Speaker 2: at people who get vaccinated later, that's not random, right,
• 14:49
  Speaker 2: So you can't infer anything about the timing of the
• 14:53
  Speaker 2: optimal timing of getting the flu shot. But what you
• 14:56
  Speaker 2: can do is say, all right, well, let's look at
• 14:57
  Speaker 2: this same group of kids we've been talking about, but
• 15:00
  Speaker 2: just focusing on the kids who have been vaccinated. And
• 15:04
  Speaker 2: we can show is that among those kids who have
• 15:06
  Speaker 2: been vaccinated, if you have a August or September birthday,
• 15:10
  Speaker 2: your vaccine tends to be given to you earlier than
• 15:14
  Speaker 2: if you have like a November or December birthday. And
• 15:16
  Speaker 2: in fact that the November December kids, they have this
• 15:19
  Speaker 2: sort of bi modal or two peak distribution. There's a
• 15:23
  Speaker 2: bunch of them who choose to get it early, but
• 15:25
  Speaker 2: there's a bunch of them who just happened to get
• 15:27
  Speaker 2: it because they showed up in their pediatrician office in
• 15:30
  Speaker 2: November and December and so what we find is that
• 15:34
  Speaker 2: the birth month with the lowest rate of influenza not
• 15:39
  Speaker 2: getting the flu shot. They all got the flu shot,
• 15:41
  Speaker 2: but the lowest rate of actually getting the flu is October.
• 15:46
  Speaker 2: And so that's kind of interesting. It tells you that
• 15:48
  Speaker 2: the optimal timing. So my chair, she was actually right
• 15:51
  Speaker 2: that I might have gotten in too early. The optimal
• 15:53
  Speaker 2: timing might be November to get the flu shot if you.
• 15:56
  Speaker 1: Have a kid. That's fascinating. The other thing that occur
• 16:00
  Speaker 1: assume you could do with this identification strategy is do
• 16:05
  Speaker 1: more on the efficacy of the flu vaccine. So there's
• 16:07
  Speaker 1: a fair amount of variation efficacy across years. Right, sometimes
• 16:11
  Speaker 1: the flu is ten percent, sometimes it's thirty percent, forty percent.
• 16:15
  Speaker 1: Usually it's higher. You could use that, right, I mean
• 16:18
  Speaker 1: you you're linking now shots to flu. You could back
• 16:22
  Speaker 1: out efficacy based on the effect sizes over time.
• 16:25
  Speaker 2: Yes, absolutely, yeah, because the you know one way to
• 16:29
  Speaker 2: do that it would be every flu season you do
• 16:31
  Speaker 2: a little randomize or a large randomized trial at the
• 16:34
  Speaker 2: very beginning of flu season to see if the thing
• 16:36
  Speaker 2: works or not. Or you could literally just use existing
• 16:39
  Speaker 2: data and in the third week of September run a
• 16:43
  Speaker 2: quick analysis to look at whether or not August born
• 16:46
  Speaker 2: kids versus September born kids have differential rates of the
• 16:50
  Speaker 2: flu shot, which we think we would see, and importantly
• 16:54
  Speaker 2: whether or not they have differential rates corresponding in influenza.
• 16:57
  Speaker 2: If they don't, that means that the flu SHOT's not
• 16:59
  Speaker 2: doing much.
• 16:59
  Speaker 1: That you could do the same I mean, I wonder
• 17:00
  Speaker 1: if the same thing is true. Now we're like so
• 17:02
  Speaker 1: in the weeds in designing our designing our research, but
• 17:05
  Speaker 1: like you could use Australia if there's the same pattern
• 17:08
  Speaker 1: in Australia since they get the Okay, all right, we
• 17:11
  Speaker 1: can work on that later.
• 17:13
  Speaker 2: Yeah, exactly, we think about the Yeah.
• 17:16
  Speaker 1: So in some sense, this chapter in this discussion I
• 17:19
  Speaker 1: think is pretty emblematic of the book. It's sort of
• 17:22
  Speaker 1: something you observed in the world in your own life,
• 17:24
  Speaker 1: and then you're thinking carefully about how to study it.
• 17:27
  Speaker 1: You know, there's a The chapter on the marathon is
• 17:30
  Speaker 1: like the one everyone is going to want to read
• 17:32
  Speaker 1: because it's super interesting and I'm not going to ruin it.
• 17:35
  Speaker 1: But that's that is a great chapter that was a
• 17:38
  Speaker 1: very clever idea and also also the cardiologist. But I
• 17:42
  Speaker 1: actually want to pivot and talk a little bit about
• 17:45
  Speaker 1: these issues of causality and and why the medical literature
• 17:51
  Speaker 1: is not good at at this often. And I'm going
• 17:54
  Speaker 1: to motivate with an example we've talked about before, which
• 17:56
  Speaker 1: is the idea of breastfeeding and links with with IQ.
• 18:01
  Speaker 1: This is not something just to be clear, this is
• 18:02
  Speaker 1: not something that's in your book, but it's on my
• 18:04
  Speaker 1: mind because there was a study published in the UK
• 18:08
  Speaker 1: a couple of weeks before we're recording this, and it
• 18:11
  Speaker 1: got a lot of attention, and it was one of
• 18:13
  Speaker 1: these many studies where they compare kids who are breastfed
• 18:15
  Speaker 1: and kids who are not, and they look at their
• 18:17
  Speaker 1: IQ scores, and there's a very deep problem with this analysis,
• 18:24
  Speaker 1: which is that the kids who are breastfed are totally
• 18:27
  Speaker 1: different on a bunch of other dimensions than the kids
• 18:30
  Speaker 1: who are not. Their parents are more educated, their moms
• 18:33
  Speaker 1: have hired cognitive scores. There's a bunch of things that
• 18:35
  Speaker 1: are that are different, and the authors aren't really able
• 18:39
  Speaker 1: to adjust for all of those, and so we would
• 18:42
  Speaker 1: sort of say this is correlation and not causation. There
• 18:47
  Speaker 1: is a lot of work like that in the medical literature.
• 18:52
  Speaker 1: You are pretty explicitly trying not to be that is that?
• 18:57
  Speaker 1: Would you say that's correct?
• 18:59
  Speaker 2: I would say, yeah, that's like I'm the anti.
• 19:01
  Speaker 1: You're like the anti that. Yeah, you're the anti that.
• 19:03
  Speaker 1: So why you know, I don't want to particularly spend
• 19:06
  Speaker 1: time dumping on this study, which was just the same
• 19:09
  Speaker 1: as all the other studies like this, but it was
• 19:12
  Speaker 1: sort of so emblematic, and I guess I just want
• 19:14
  Speaker 1: to understand, like what is your sense of of why,
• 19:17
  Speaker 1: like why is there not more BAPU? Why is there
• 19:19
  Speaker 1: too much of this other thing and not more and
• 19:22
  Speaker 1: not more BAPU style research?
• 19:25
  Speaker 2: First I asked my wife that the state, why is
• 19:27
  Speaker 2: it there? More about By the way, just to be clear,
• 19:30
  Speaker 2: are you prepared to state that peanuts do not cause dementia?
• 19:33
  Speaker 2: Just to be clear, I'm I'm prepared.
• 19:35
  Speaker 1: I'm prepared to state that chia seeds do not prevent
• 19:38
  Speaker 1: it and peanuts don't cause it. I'm ready. We'll be up, coach.
• 19:41
  Speaker 2: I think, I mean, it's it's a great question, and
• 19:47
  Speaker 2: probably I joke about peanuts and dementia, but there we
• 19:50
  Speaker 2: see this problem. Probably the most is in nutritional epidemiology,
• 19:55
  Speaker 2: and I think it's probably two things. One is that
• 19:58
  Speaker 2: there may be a generalized lack of awareness of how
• 20:02
  Speaker 2: serious these problems are and how people actually look at
• 20:07
  Speaker 2: these sorts of evidence and try to make decisions on that.
• 20:09
  Speaker 2: And it's not just a matter of publishing research. People
• 20:11
  Speaker 2: actually listen to this stuff, so you've got to be cautious.
• 20:14
  Speaker 2: So there's probably a generalized lack of appreciation of how
• 20:17
  Speaker 2: serious an empirical issue this is, and that's certainly true
• 20:22
  Speaker 2: I think in medicine. The second thing is that I
• 20:27
  Speaker 2: think that there's an incentives problem that we have here.
• 20:30
  Speaker 2: What I mean by that, well, think about where is
• 20:32
  Speaker 2: this stuff coming from. This is research, right, These are
• 20:34
  Speaker 2: researchers who are publishing research that is being published in
• 20:39
  Speaker 2: research journals. And my view is that if the journals
• 20:43
  Speaker 2: sort of celebrate this type of work, they publish relationships
• 20:48
  Speaker 2: between exercise and mortality, or coffee consumption and cancer, whatever
• 20:53
  Speaker 2: it may be, if they reward authors with that kind
• 20:56
  Speaker 2: of performing that kind of work by publishing it, why
• 21:00
  Speaker 2: wouldn't we expect to see this Like I in some sense,
• 21:02
  Speaker 2: I think that researchers are probably responding to the incentives
• 21:05
  Speaker 2: that they have available to them. So take the take
• 21:08
  Speaker 2: the converse in economics. In economics, do we not see
• 21:12
  Speaker 2: this sort of poorly thought out research because economists are
• 21:16
  Speaker 2: better trained? Yeah, they probably are better trained. But it's
• 21:19
  Speaker 2: also the fact that you will never be able to
• 21:21
  Speaker 2: get these kinds of studies into economic journals. It's just
• 21:24
  Speaker 2: not going to fly. So there's no incentive to try
• 21:26
  Speaker 2: to do that.
• 21:27
  Speaker 1: Yeah, I mean, we have our own problems, but I
• 21:28
  Speaker 1: agree this particular issue is not is not one that
• 21:32
  Speaker 1: that's true.
• 21:33
  Speaker 2: Yes, there's other problems that, Yeah, that's true.
• 21:35
  Speaker 1: This is the thing. But it's interesting because I have
• 21:38
  Speaker 1: the same instinct. But I pose this question to my
• 21:41
  Speaker 1: class once and we basically got there and then kept
• 21:45
  Speaker 1: going down the rabbit hole. So think about, like, what's
• 21:48
  Speaker 1: the incentive for the journal? So why is the journal
• 21:51
  Speaker 1: publishing this? And in my view, the answer is like,
• 21:53
  Speaker 1: why is the journal publishing this? Because it's going to
• 21:56
  Speaker 1: be covered in the media. I mean the journal. One
• 22:00
  Speaker 1: part of the journal's incentive is attention, right, and media
• 22:05
  Speaker 1: loves you know, peanuts caused dementia? Right, Peanuts caused dementia
• 22:08
  Speaker 1: is a great headline. You know, she sees don't cause
• 22:11
  Speaker 1: demension coffee does?
• 22:12
  Speaker 2: Is?
• 22:12
  Speaker 1: People love that stuff? They like to read about it,
• 22:15
  Speaker 1: and so that is the incentive for the media. Then
• 22:18
  Speaker 1: that is the incentive for the journal. It's actually not
• 22:21
  Speaker 1: that interesting to publish a paper that says there's no
• 22:23
  Speaker 1: link between any known foods and dementia, which more or
• 22:27
  Speaker 1: less true. That's not a very interesting journal title, and
• 22:31
  Speaker 1: it's not getting you on the BBC or in the
• 22:33
  Speaker 1: New York Times. Then when I pose that, then my
• 22:36
  Speaker 1: students were like, well, the reason that the journal why
• 22:38
  Speaker 1: are the newspapers publishing this? Well, why are the newspapers
• 22:41
  Speaker 1: publishing a PAPU?
• 22:43
  Speaker 2: Yeah, I mean I think they'd publish it, but because
• 22:46
  Speaker 2: they want clicks. But I think you could go further
• 22:48
  Speaker 2: into yeah.
• 22:48
  Speaker 1: Yeah, so ultimately it's your fault. It's our fault for clicking.
• 22:52
  Speaker 1: That was where we got to students were the student
• 22:54
  Speaker 1: was like at the end, he was like, actually, I
• 22:56
  Speaker 1: think it's my fault because I read that article like
• 22:59
  Speaker 1: I'm incentivizing them.
• 23:01
  Speaker 2: Yeah, but I would say it's not our fault in
• 23:03
  Speaker 2: the following sense. We click that because we want to
• 23:07
  Speaker 2: know the answer to a question that matters to us.
• 23:10
  Speaker 2: But we're tricked because what we're clicking we don't know,
• 23:14
  Speaker 2: doesn't answer the question that matters to us. It's just
• 23:17
  Speaker 2: purports to do that. And I think that's where the
• 23:19
  Speaker 2: gatekeepers need to be involved. And to be clear, I
• 23:22
  Speaker 2: don't actually think that The New York Times is at
• 23:24
  Speaker 2: fault here, Like I don't expect a really well trained
• 23:28
  Speaker 2: journalist to be able to parse out these issues. This
• 23:29
  Speaker 2: is something that the journals should be the gatekeeper for,
• 23:33
  Speaker 2: but for whatever reason they're not.
• 23:36
  Speaker 1: Yeah, I know, I think that's right, and I think
• 23:37
  Speaker 1: it is on the there's a sense in which maybe
• 23:42
  Speaker 1: some of the problems are really in the training and
• 23:46
  Speaker 1: in people's understanding. And you know, I said that I
• 23:49
  Speaker 1: thought it was obvious that some of these relationships are
• 23:52
  Speaker 1: just correlation and not causality. I do think there's a
• 23:55
  Speaker 1: perspective that you and I shared that is perhaps not
• 23:58
  Speaker 1: universally shared, that these problems associated with comparing two groups
• 24:04
  Speaker 1: with some adjustments. I think we have a sense that
• 24:07
  Speaker 1: like that's more or less it is more or less
• 24:10
  Speaker 1: impossible to learn about the impacts of diet or many
• 24:14
  Speaker 1: other things from that type of analysis. But not everyone
• 24:17
  Speaker 1: would agree with that.
• 24:18
  Speaker 2: That's right, Yeah, And I mean I've thought a lot
• 24:21
  Speaker 2: about not very productively, but you know, I've thought a
• 24:24
  Speaker 2: lot about how do we get at these questions of
• 24:27
  Speaker 2: the impact of diet using some of the tools that
• 24:30
  Speaker 2: we like, these natural experiments, And it's hard. I mean,
• 24:33
  Speaker 2: I remember when you came to my class a couple
• 24:35
  Speaker 2: of months ago. I got really excited about an idea
• 24:38
  Speaker 2: that you basically shot down. Wasn't that it wasn't good?
• 24:41
  Speaker 2: I mean, it wasn't great. I don't know if it
• 24:42
  Speaker 2: wasn't Wasn't.
• 24:43
  Speaker 1: It good the idea earlier in.
• 24:44
  Speaker 2: This podcast that's true?
• 24:47
  Speaker 1: Yeah, no, But I mean I think that for me
• 24:50
  Speaker 1: that illustrates one limitation with the approach that you take
• 24:54
  Speaker 1: in all of these in all of the chapters really
• 24:57
  Speaker 1: in the book, which is that they're important, but the
• 25:01
  Speaker 1: set of questions that you're able to answer here are limited.
• 25:04
  Speaker 1: And if you started with you know, I want to
• 25:06
  Speaker 1: understand the impact of axon hy or you know, the
• 25:09
  Speaker 1: impact of this kind of diet on outcomes, it's that's
• 25:14
  Speaker 1: a really hard question to get at with the kinds
• 25:17
  Speaker 1: of experiments, the kinds of sort of naturally occurring experiments
• 25:21
  Speaker 1: that you have here. And so in a lot of
• 25:23
  Speaker 1: places you're taking advantage of something where nature has randomized
• 25:26
  Speaker 1: it for you, But there are many things that nature
• 25:29
  Speaker 1: doesn't randomize for us. And then there's a temptation to say, well,
• 25:33
  Speaker 1: I guess the only way we can learn about this
• 25:35
  Speaker 1: is with this more of something that smacks more of correlation.
• 25:39
  Speaker 2: Yeah, yeah, that's right.
• 25:41
  Speaker 1: Yeah.
• 25:41
  Speaker 2: I mean I was thinking of the other day I
• 25:44
  Speaker 2: almost packed the peanut butter sandwich for one of our
• 25:47
  Speaker 2: kids going to school, and I realized I wasn't allowed
• 25:51
  Speaker 2: to do that, but it made me think. I was like, wow,
• 25:52
  Speaker 2: what you know, there was a point in time in
• 25:54
  Speaker 2: which that would have been okay, But there might be
• 25:57
  Speaker 2: some classes where there is a child with a peanut
• 25:59
  Speaker 2: allergy just by chance, right, and my child who does
• 26:03
  Speaker 2: not have a pen and aalogay would not have selected
• 26:05
  Speaker 2: into his classmate's class based on whether or not his
• 26:09
  Speaker 2: classmate had a pen analogy. It'd be totally random. So
• 26:12
  Speaker 2: might you use that sort of variation to figure out
• 26:15
  Speaker 2: what is the causal effect of differential amounts of peanut exposure?
• 26:19
  Speaker 2: But the problem that doesn't work is because that's just
• 26:21
  Speaker 2: one one iota of a place where my son would
• 26:26
  Speaker 2: be exposed to peanuts. So yeah, it's random, but it's
• 26:29
  Speaker 2: not going to mass.
• 26:29
  Speaker 1: So I thought you were going to say something different
• 26:31
  Speaker 1: about peanut exposure, which is this, for me, is the
• 26:34
  Speaker 1: strongest example of where we can use correlational data to
• 26:38
  Speaker 1: motivate us. So it used to be that people were told,
• 26:41
  Speaker 1: you know what, my kid, my first kid was born.
• 26:44
  Speaker 1: You were told, don't give your kids peanuts before they're
• 26:46
  Speaker 1: too because that will they will develop an allergy, Like
• 26:49
  Speaker 1: they'll be more likely to become allergic. There's a guy
• 26:52
  Speaker 1: named Gideon Lack in the UK and he did this
• 26:56
  Speaker 1: very like what you would see like classic correlational experiment,
• 27:00
  Speaker 1: comparing kids in the UK and kids in Israel and
• 27:03
  Speaker 1: showing kids in Israel are less likely to have peanut allergies.
• 27:07
  Speaker 1: And he then concluded that that was because they were
• 27:10
  Speaker 1: eating this peanut snack when they were like four months old.
• 27:12
  Speaker 1: So he like had this is like the thing we
• 27:14
  Speaker 1: would just be like, are you kidding me? You compared
• 27:17
  Speaker 1: kids in the UK and Israel and then you decided
• 27:19
  Speaker 1: it was because of this peanuts snack, Like you're a like,
• 27:22
  Speaker 1: this is so much worse than all of the it's
• 27:25
  Speaker 1: like just seems like a ridiculous method. But then he
• 27:28
  Speaker 1: did a randomized trial and it turns out that exposing
• 27:30
  Speaker 1: your kids to peanuts and they're like four to six
• 27:32
  Speaker 1: months actually lowers the risk of developing an allergy by
• 27:35
  Speaker 1: seventy percent. So it turns out he was totally right.
• 27:39
  Speaker 1: And but it is. It's an example if you say, like,
• 27:42
  Speaker 1: what's the observational data good for It's good for hypothesis generation,
• 27:45
  Speaker 1: and you see that, but it's not enough. You wouldn't
• 27:48
  Speaker 1: want to change the way you were doing, you know,
• 27:50
  Speaker 1: peanut exposure for kids based on his observation about the
• 27:54
  Speaker 1: Israeli kids. But maybe you do want to change, you know,
• 27:57
  Speaker 1: how you do your research. And once, of course, you
• 27:59
  Speaker 1: have a randomized trial like that, then there's a lot
• 28:01
  Speaker 1: of reason to do it. So anyway, that's mine. That's
• 28:04
  Speaker 1: where I thought you're went, Yeah.
• 28:05
  Speaker 2: You know, okay, well let me tell you where I
• 28:08
  Speaker 2: was going to go. Just now. You know, I like
• 28:09
  Speaker 2: to play soccer during lunchtime. That's one of the benefits
• 28:11
  Speaker 2: of this academic lifestyle. And there's a phrase that always
• 28:15
  Speaker 2: gets repeated to me anytime I happen to score or
• 28:18
  Speaker 2: do something right. There's a guy who just says, bapu,
• 28:21
  Speaker 2: a broken clock is right two times a day. H Okay, great, Thanks,
• 28:25
  Speaker 2: Thanks dude. I appreciate that you know, you can stumble
• 28:27
  Speaker 2: upon by accident.
• 28:31
  Speaker 1: Yeah, how often would you say you score? Are you
• 28:33
  Speaker 1: a good good soccer player? No?
• 28:36
  Speaker 2: No, no, I'm not very good now, but I try hard.
• 28:38
  Speaker 2: That's all that matters.
• 28:39
  Speaker 1: I feel like, you know, we're getting we're about the
• 28:41
  Speaker 1: same age maybe, And I feel like over time, if
• 28:44
  Speaker 1: you can just keep playing, like eventually, people are just
• 28:47
  Speaker 1: going to kind of peet and their knees are going
• 28:49
  Speaker 1: to go so pretty much, if you keep going, eventually
• 28:52
  Speaker 1: you'll be good.
• 28:53
  Speaker 2: That's right.
• 28:53
  Speaker 1: Yeah, That's how I'm working with running. I'm just like,
• 28:56
  Speaker 1: I'm going to keep going and people are just aging.
• 28:58
  Speaker 1: You know, I just got to keep going.
• 29:00
  Speaker 2: But yeah, this is like a great question. Like, for example,
• 29:03
  Speaker 2: if you want to think about the effects of exercise
• 29:06
  Speaker 2: on health, what type of exercise is optimal? You can
• 29:09
  Speaker 2: do a randomized trial where you encourage people to run
• 29:11
  Speaker 2: wrong distances, or you could do sort of high intensity workouts.
• 29:14
  Speaker 2: But I think are there examples of natural experiments that
• 29:18
  Speaker 2: get people to do certain types of exercises versus others
• 29:21
  Speaker 2: in a way that could tell us something about the
• 29:23
  Speaker 2: effects of that type of exercise on health, and it's
• 29:27
  Speaker 2: you know, really hard to find those sort of examples.
• 29:30
  Speaker 1: I agree. I mean exercise an interesting one where we
• 29:33
  Speaker 1: do have some randomized trials, but it's mostly on things
• 29:35
  Speaker 1: like do you exercise at all? You know, do you
• 29:39
  Speaker 1: walk after lunch? You know, do you do any kind
• 29:42
  Speaker 1: of aerobic exercise? But then when you get into these
• 29:45
  Speaker 1: more nuanced things like is it better to play soccer?
• 29:47
  Speaker 1: Is it better to run? Is it better to swim?
• 29:49
  Speaker 1: Like is it better to lift? You do CrossFit, then
• 29:52
  Speaker 1: it's harder to imagine the experiment.
• 29:55
  Speaker 2: Yeah. The other thing that I think we don't really
• 29:58
  Speaker 2: appreciate much. This is of a methodological point, but I
• 30:01
  Speaker 2: think it's important for trying to understand what causes what
• 30:05
  Speaker 2: You imagine that you randomize people to exercise versus not.
• 30:09
  Speaker 2: And so you'd have that randomization, that would be great,
• 30:12
  Speaker 2: But what happens if the people who exercise start doing
• 30:16
  Speaker 2: other things because they are exercising that counteract the benefits
• 30:20
  Speaker 2: of exercise. So, for example, when I exercise a lot,
• 30:23
  Speaker 2: I will often go and eat out for lunch because
• 30:25
  Speaker 2: I feel like I earned it. And if you see
• 30:28
  Speaker 2: that sort of phenomenon happening even after randomization, you don't
• 30:31
  Speaker 2: actually know what the cause of effect of the exercise
• 30:34
  Speaker 2: is because the behaviors are modified as a result of
• 30:37
  Speaker 2: that treatment. So I think that's a thorny problem to solve.
• 30:40
  Speaker 1: That comes up even, I mean that comes up and
• 30:42
  Speaker 1: we look at something like statins, right, like it's cholesterol
• 30:45
  Speaker 1: lowering medications. If you're taking the cholesterol lowering mecation also
• 30:48
  Speaker 1: as a result thinking well, I can just you know,
• 30:51
  Speaker 1: eat a bunch of I can eat more cholesterol because
• 30:53
  Speaker 1: I have this backstop than you know your your effects are.
• 30:58
  Speaker 1: That's an interesting method a thing because when the experiments
• 31:01
  Speaker 1: in the randomized experience where they evaluate the effects of statins,
• 31:05
  Speaker 1: they you don't know that you're taking it, so you
• 31:07
  Speaker 1: don't see as much of that. Then when it enters
• 31:09
  Speaker 1: the real world, you may expect the effects to be
• 31:11
  Speaker 1: smaller because people are compensating because now they know that
• 31:14
  Speaker 1: they're taking the drug.
• 31:16
  Speaker 2: Exactly, yeah, exactly, that's right.
• 31:17
  Speaker 1: Yeah, all right. To finish, I'm sure we could do
• 31:20
  Speaker 1: this for the whole day, but we've already come up
• 31:23
  Speaker 1: with one paper idea, So I think that's pretty productive
• 31:26
  Speaker 1: at least.
• 31:26
  Speaker 2: Yeah, but let me just be clear, if this were
• 31:28
  Speaker 2: a medical journal, we could take that paper and spit
• 31:30
  Speaker 2: it into six ideas.
• 31:33
  Speaker 1: I wait, I can't poop where I eat, So tell
• 31:38
  Speaker 1: me what is your favorite chapter in the book?
• 31:41
  Speaker 2: My favorite chapter?
• 31:43
  Speaker 1: You know what?
• 31:43
  Speaker 2: These are like babies. I love them all. Yeah, I
• 31:46
  Speaker 2: love them all so much. I kind of like the
• 31:51
  Speaker 2: last chapter, which is about dealt with issues around COVID nineteen.
• 31:56
  Speaker 2: I mean, to be fair, I like all the chapters,
• 32:00
  Speaker 2: but I'll pinpoint this particular chapter because it does relate
• 32:04
  Speaker 2: to something else in our conversation, which is, when COVID
• 32:08
  Speaker 2: nineteen started, there was this huge influx, as you know,
• 32:13
  Speaker 2: of research, some high quality, much of it very low quality,
• 32:17
  Speaker 2: and I sort of made an intentional decision early on
• 32:21
  Speaker 2: not to work on any topics related to COVID nineteen
• 32:24
  Speaker 2: unless I thought I could do something different, And so
• 32:27
  Speaker 2: as a result, I think I maybe had one COVID
• 32:29
  Speaker 2: nineteen paper, which is in the book, But it sort
• 32:32
  Speaker 2: of speaks to the incentives. I mean, it's true, probably
• 32:35
  Speaker 2: people were interested in solving problems and expanding knowledge, but
• 32:39
  Speaker 2: at the same time, there's so many bad papers out
• 32:41
  Speaker 2: there by people who are smart and presumably have some
• 32:44
  Speaker 2: methodological training, and you kind of wonder why they even
• 32:47
  Speaker 2: thought to do that in the way they did. But
• 32:50
  Speaker 2: the one paper I had with Chris Whaley and a
• 32:54
  Speaker 2: couple other people is we looked we were interested in
• 32:57
  Speaker 2: this question of well, actually, let me tell you what
• 33:01
  Speaker 2: the paper was. And then I backtracked what the motivation
• 33:03
  Speaker 2: was after I had the idea. So basically, I was like,
• 33:06
  Speaker 2: all right, I'm looking at this insurance data one day
• 33:09
  Speaker 2: and I'm recognizing that we have information on when people
• 33:11
  Speaker 2: have birthdays. And it was around the time of our
• 33:15
  Speaker 2: daughter's birthday, and we had to make a decision whether
• 33:17
  Speaker 2: or not to host it via zoom or in person,
• 33:21
  Speaker 2: and we made a decision to do it via zoom
• 33:24
  Speaker 2: with this magician who was quite phenomenal. And so I
• 33:28
  Speaker 2: was looking at the data one day and I was like, oh, wait, wait, wait,
• 33:31
  Speaker 2: this actually relates to this decision that my wife and
• 33:33
  Speaker 2: I had to make. I bet that people are more
• 33:37
  Speaker 2: likely to get COVID nineteen after their birthday because they
• 33:41
  Speaker 2: might celebrate with other people, be exposed to a larger
• 33:44
  Speaker 2: group than they otherwise would be, even if they were
• 33:46
  Speaker 2: being cautious, And might this be an interesting question to study?
• 33:50
  Speaker 2: And at that point I didn't sort of have a
• 33:51
  Speaker 2: motivation how to sort of sell or motivate the paper.
• 33:54
  Speaker 2: So we looked at the data and you see very
• 33:57
  Speaker 2: clearly that if you look within the same city, within
• 34:00
  Speaker 2: the same let's say week of the year, households in
• 34:02
  Speaker 2: which someone has a birthday, those households are nearly identical
• 34:07
  Speaker 2: to households in which no one has a birthday that
• 34:10
  Speaker 2: week in that same city. But the first set of
• 34:13
  Speaker 2: households are about twenty percent more likely to get a
• 34:15
  Speaker 2: new COVID nineteen diagnosis in the next two to three weeks.
• 34:19
  Speaker 2: And that to me was like, oh, that's an interesting finding.
• 34:22
  Speaker 2: And then I was sort of tackling the problem with
• 34:25
  Speaker 2: my colleagues, well, how do we quote unquote sell it,
• 34:28
  Speaker 2: how do we motivate what the question is? And then
• 34:31
  Speaker 2: the motivation I think we settled on was, well, look,
• 34:33
  Speaker 2: early in the pandemic, there's all this discussion about super
• 34:37
  Speaker 2: spreader events being a common source for spreading COVID nineteen
• 34:42
  Speaker 2: the virus, and we didn't really have an appreciation of
• 34:45
  Speaker 2: whether or not small gatherings with people that you know
• 34:48
  Speaker 2: and trust. It's not like going to a random place
• 34:51
  Speaker 2: like a bar, but people that you know and you trust,
• 34:53
  Speaker 2: you think that they know how the living their lives,
• 34:56
  Speaker 2: whether those sorts of gatherings could have an impact. And
• 34:59
  Speaker 2: the first thing we found is yes, they do have
• 35:00
  Speaker 2: an impact. And then the second thing that was interesting
• 35:03
  Speaker 2: to me, and this is maybe a broader reflection of
• 35:07
  Speaker 2: some of the issues that you've been thinking about and
• 35:08
  Speaker 2: writing about a lot during the pandemic, was we found
• 35:11
  Speaker 2: that this birthday effect was basically the same in really
• 35:15
  Speaker 2: republican areas and really democratic areas. And you might think
• 35:19
  Speaker 2: that if Democrats were much more conservative about social distancing
• 35:23
  Speaker 2: that kind of stuff, that there would be no quote
• 35:25
  Speaker 2: unquote birthda effect there, but it was the exact same,
• 35:28
  Speaker 2: and it made me reflect on this thing that we
• 35:30
  Speaker 2: always think about economics is that you got to look
• 35:32
  Speaker 2: at people's behaviors. You can't look at what they say
• 35:34
  Speaker 2: they do. You got to look at what they do.
• 35:36
  Speaker 2: And here's what they did. They get together. And by
• 35:40
  Speaker 2: the way, the effect is larger when it's a kid's
• 35:43
  Speaker 2: birthday than an adult birthday because partly.
• 35:46
  Speaker 1: Because adults are not you know, they don't they don't
• 35:49
  Speaker 1: have as good birthday parties. That's the message. Yeah exactly, Yeah,
• 35:56
  Speaker 1: thank you so much. It was a tree to talk.
• 35:58
  Speaker 1: The book is fantastic and everyone should pick it up
• 36:02
  Speaker 1: and read it. I appreciate it.
• 36:04
  Speaker 2: Thank you, it's been fun.
• 36:11
  Speaker 1: Thanks for listening. If you like what you heard, subscribe
• 36:15
  Speaker 1: to parent Data in your favorite podcast app and rate
• 36:18
  Speaker 1: and review the show in Apple Podcasts. You can subscribe
• 36:21
  Speaker 1: to the whole newsletter for free at www dot parentdata
• 36:25
  Speaker 1: dot org. Talk to you soon.