An Oddly Timely Interview (also, hi!)
I'm gonna ignore the fact that it's been a while and jump right in. For an almost too-perfect last piece before I go on maternity leave, I interviewed Emily Oster, author of Expecting Better and the new Cribsheet. Expecting Better was like a talisman through my pregnancy, and it was honestly a special experience to get to talk to Oster at this moment in my life. Our interview is up today on The Cut, but there was so much that couldn't fit into that space. Here are some of my favorite extras:
On that op-ed and interpreting studies:
A few months ago there was an op ed in the New York Times by the obstetrician Jen Gunter, who said that no alcohol during pregnancy is safe. Before I ask you about the actual science of it, I want to ask about how that author framed the relationship between pregnant people and their doctors. She writes that a less restrictive pregnancy — an occasional drink, eating some sushi — may feel like a strike against the patriarchy. But then she says that the science doesn’t support that in terms of drinking. Do you think that there is an element of patriarchy or oppression in the way that pregnant people are treated by medicine, or are we just dealing with the practical limits of research?
I think the main problem is research quality or quantity, or the ability to do research, not the patriarchy. There is this infantilism aspect of some interactions during pregnancy, which a lot of women experience. I'm not sure how gendered that is — I think that there are aspects of interaction with the medical system that people of all genders find frustrating in some of the same ways. So I'm not sure that I would go all the way to the patriarchy explanation, but I think that's also a question of lived experience.
So then in terms of the varying qualities of the research and data that we have about all of these questions in pregnancy, do you have a take on the scientific argument of that op ed?
The best evidence we have, [which] compares women who have had a small amount of alcohol to women who have not, and looks at the outcomes for their kids, is basically reassuring [that an occasional drink is safe]. [The research cited in] this article looks for kids who they think have been affected by Fetal Alcohol Spectrum Disorder, or Fetal Alcohol Syndrome. It just isn't a study about the effect of having an occasional drink.
How would you hope that a lay reader could handle encountering an op ed like that or other write ups of new research?
There are some things that make a study "more compelling." Things like, is this a randomized trial? Or, does this study do a good job adjusting for other differences across people? Is it a big study?
I think there were two things missing from that op ed which would be very hard for laypeople to evaluate. One is,"Where does this study fit inside the other literature?" And two, in this example, this op ed was about a study which was itself a subject of a lot of media coverage, but a while ago, so you would have had to dig into the media coverage from a year ago to figure out what some of the issues were.
On Oster's academic career:
You obviously rely on your academic background to be able to parse these studies, because while you’re not a medical researcher, your work in economics is very much related. What is your academic research on?
I study health behaviors, and I do some method [analysis]. The particular topics of my research are not super relevant to the books for the most part, but two things that are really relevant are: the general science of decision-making that is part of economics overall; and the emphasis on causality, and trying to understand which relationships [in data] are causal and which are not, which is quite relevant for evaluating the types of studies I talk about in the books.
On different ways to make decisions from data and avoiding confirmation bias (i.e. let's go to econ class):
I want to get into one nitty gritty question, about Bayesian priors. But first, so that I don't have to do it in my writing, could you explain how a Bayesian approach to interpreting data works?
When we look at evidence, the “frequentist” way to do it is you look at the evidence you have in front of you and you draw a conclusion based only on [that]. Then there's an approach people call Bayesian, where you start with some existing belief, your “prior,” and you use any given piece of evidence to “update” your prior, to incorporate it along with what you thought before.
So if you were very sure that something was good for your kid before, and then a study comes out that says that it's bad, a frequentist might immediately be like, "Okay, now that this latest study says that it's bad, I'm sure that this thing bad." A Bayesian is going to say, "Well I thought it was good before, so maybe this moves me to think it's a little bit less good than I did before, but because my prior is very strong, this isn't gonna move me all the way to thinking it's bad."
How is that different from confirmation bias? In the section on vaccination you highlight these inconclusive or weak studies that allow parents on either side of the question to find confirmation for their preconceptions. But later you write about Bayesian decision making as a useful way of making decisions. How do those differ?
Confirmation bias refers to a situation where you interpret new data through the lens of a bias. So rather than using data for what it shows, you’re going to twist it to assume that it supports what you thought before. The Bayesian approach isn’t taking the data and interpreting it funny—the data should move everybody in the same direction, but it’s going to move you differently based on where you started.
If you or someone in your life is pregnant or raising small kids, I sincerely can't recommend Oster's books enough.
As for me and what's coming, this tinyletter has flourished when I've had too much on my mind and nowhere to put it... but obviously pregnancy was a fallow period. So basically I have no idea what will happen next—in terms of this tinyletter or at all! I can at the very least try to be better about keeping you updated with what I'm working on. A couple of things I've written since the last letter I sent:
I wrote about how I changed my skincare routine in pregnancy for The Cut, entirely against any data or scientific necessity but because it felt nice.
I took advantage of Valentine's Day to write about romance novels for Vulture, specifically about why you, readers who do not already read romance novels, should.
I also was on Bobby Finger's podcast, I'm Obsessed With This, to talk about the TV shows I love for being even more soothing than Bake-Off.
Oh also I sold a book! Based on my Medium column from last year. I still have to write the book, so keep an eye out for it in... 2021?
That's it for now. See you on the other side.
Jaime