This is a good article by Emily Oster, an economist, on pregnancy in the WSJ. She takes a closer look at the research behind the dos and don'ts that overwhelm expecting mothers. I have a mix of reactions here.
First, this is another example (like Bob Murphy's discussion the other day of military suicides), where experimental scientists are often at a disadvantage in interpreting non-experimental data relative to economists. I have to trust Oster's read of the studies because I haven't looked through them myself, but it sounds like the issue of endogeneity is completely absent - particularly in some of the most widely cited studies. Again, this shouldn't be surprising. Medical research is principally experimental research, so when a future medical researcher is sitting through stats classes they're not going to spend all the time that economists do talking about endogeneity and model identification.
When you look at studies that do a better job dealing with this the results are less scary - moderate drinking and caffeine intake don't seem to have an identifiable impact.
The other thing she does that I like is that she tries to get a better sense of the probability distributions underlying the dos and don'ts. This is extremely frustrating around weight issues where I'd agree that a lot of doctors act like there is not a continuous distribution of expected weight gains - that if you're one pound over the expected gain it's a problem rather than a marginal increase in the context of a probability distribution with a high level of variability in the first place. I also liked her more detailed look at exactly what foods pose a listeria risk (nasty stuff that pregnant women are far more susceptible to and can cause blindness in the baby). Apparently deli turkey is a real problem for it... but deli ham is very low risk. You wouldn't know that from the doctors - they just tell you to stay away from all deli meats.
I do have to disagree with Oster and people who responded to her like Art Carden on one issue. Art writes: "Ms. Oster's essay speaks to modernity's outsized obsession with minor risks generated by a news cycle that is constantly warning us about monsters under the bed or in the coffee pot. Fretting about trivial risks is itself risky; it leaves us with less time and attention to deal with more substantial concerns."
When you're pregnant you have a lot on your mind and a lot to do in a short amount of time. I imagine this is particularly true of first-timers. Although as a data guy I do of course think it's good to know what's behind these dos and don'ts and make decisions accordingly, I also think there's a good reason for the more straightforward rules. Behavioral economics tells us that people use heuristics because we are working with limited cognitive resources. We have a lot on our plate, particularly those of us expecting a baby. Kate and I like wine a lot and Kate knows that a little wine is not going to be a problem for the baby, but she's refused to drink (and she has one cup of coffee a day) through the pregnancy because it's much easier to do that than it is to weigh all the competing risks and keep track of how much she drank this week. It's a heuristic. It lets her bracket off that whole issue, not think about it any more, and focus on navigating the nursery, work, the sequester, the doctors visits, and the pregnancy classes. It's a stress reducer. If she continued drinking she would constantly be second guessing any given decision to have a drink and she'd be miserable.
It would be nice if doctors could provide more detail on the data associated with all this but they have limited time too and I don't think they present the simple dos and don'ts to make people worry. Quite the opposite - they do it precisely so people don't have to fret over it.
For the Weekend...
6 hours ago