Imagine that the President convenes his top economic advisors to get a recommendation on a pressing policy issue. They say unequivocally “do X.” The President asks why and they say “its complicated. Do X.” The President, not happy with that, decides he is going to read the economic literature on the pros and cons of doing X. After a thorough study he comes back to his advisors and says “You economists don’t understand your own science. I read the literature and I should do Y.”
I think we would agree that’s a bad outcome. For probably exactly the same reason that Doctors don’t seem to be happy with economist Emily Oster’s apparent advice to pregnant women to drink alcohol “like a European adult.”
But let’s assume that Emily truly can interpret the published statistical literature better than her Obstetrician. There is another reason to question her recommendations.
An advisor’s job is to advise on the risks of an activity. Because the advisor is the expert on that. The decision-maker is the expert on her own preferences. The correct decision is based on weighing both of these.
A recommendation to have up to a glass of wine per day while pregnant confounds the two sides. What it really means is “I like wine a lot. I also read about the risks and decided that my taste for wine was strong enough that I am willing to live with the risks.” Thus her recommendation amounts to “If you like wine as much as I do you should drink up to a glass per day when you are pregnant.”
When I asked my doctor about drinking wine, she said that one or two glasses a week was “probably fine.” But “probably fine” isn’t a number.
The problem is that there is no way to quantify how much she likes wine and so no way for her readers to know whether they like wine as much as she does. Likewise it is too much for Emily to demand her doctors to say much more than “probably fine.”
The doctors’ advice is based on some assumption about the patient’s taste for wine weighed against the risks. Emily’s advice is based on a different assumption. As for the risks, when Emily reads the literature and concludes that the evidence is weak of the danger of drinking alcohol she then jumps to the conclusion that it is weaker than what the doctors thought. She makes the identifying assumption that their recommendation was conservative because they overestimated the risks and not because they underestimated her taste for wine. But there does not seem to be any basis for that assumption because her doctors never told her what they believed the risks to be and they never asked her how much she likes wine.
Cheap Talk 
12 comments
Comments feed for this article
September 11, 2013 at 6:51 am
O. S.
One reason I don’t like Emily Oster is that the she does not seem to take into account that she is not your average mother in the US, or for that matter any country in the world. The advice she dispenses in the book may be fine for her and others with similar socio-economic characteristics but not for many others. Take some poor single mother, with limited education and low social capital. In all likelihood, such an individual has poor decision-making skills due partly due to their discount rate. Any advice to such an individual that does not explicitly discourage the consumption of alcohol during pregnancy is irresponsible. There is something to be said for paternalism sometimes, especially when the vast majority of people who will be starting families are without advanced social science degrees that enable them to shift through volumes of research.
September 11, 2013 at 8:46 am
dan s
the fact that doctors ‘never asked her how much she likes wine’ could be key. they make the same advice for those who love wine and those who don’t. one could argue this is sufficient to imply doctor overestimation of alcohol risk
September 11, 2013 at 9:57 am
E
^ By Jensen, if doctors are risk averse…
September 11, 2013 at 2:25 pm
dan s
yeah if risk averse and they couldn’t get info about wine preferences – then maybe should just advise avoiding wine.
but they could get this info, just by asking. (well i realize there could be issues here but seem second order). and given that doctors could ask but don’t, that could mean they think no one should be drinking wine (even wine lovers)
September 11, 2013 at 2:33 pm
jeff
Another way of saying things is this. In her WSJ piece she complained that her doctor told her that 1 glass of wine per week is “probably fine.” In particular she complained that “probably fine” is not a number.
I liked that comment because it points to what an economist can truly add to the discussion. Patients need to know the tradeoffs.
The problem is, after she went through the data she didn’t come up with a number either. As she wrote in her Slate article (entitled “I Wrote That It’s OK to Drink While Pregnant. Everyone Freaked Out. Here’s Why I’m Right”) mothers should “feel comfortable” having a glass of wine per day.
“feel comfortable” is not a number. but of course there was never any hope of coming up with that number. in the end she did exactly what her doctor did: formed a belief about the risks, made an assumption about the utility of drinking wine and announced to the world the optimal policy under those two realizations of the variables.
from that announcement we cannot infer (her beliefs about) the risks because of the identification problem. just like from her doctor’s recommendation she could not infer the risks.
March 20, 2014 at 9:54 pm
Open
07-01-09d094d0b8d0bcd0b0d181d182d18bd0b9 d0b8d0b7 d0bcd0b0d0b3d0b0d0b4d0b0d0bdd0b0 =)) d0bfd0b8d188d0b5d182: d18f d0bfd0be d0bdd0b0d187d0b0d0bbd183 d182d0bed0b6d0b5 d0bdd0b5 d181d0bcd0bed0b3 d180d183d181d0b8d184d0b8d186d0b8d180d0bed0b2d0b0d182d18c, d0b7d0b0d0bad0b8d0bdd183d0bb d184d0b0d0b9d0bb d0b2 d0bad0bed180d0bdd0b5d0b2d0bed0b9 d0bad0b0d182d0b0d0bbd0bed0b3 d0b8 d0bdd0b8 d187d0b5d0b3d0be d0bdd0b5 d0bfd180d0bed0b8d0b7d0bed188d0bbd0be, d0bdd0b5 d0bed182d0bed0b1d180d0b0d0b7d0b8d0bb d0bed0bd d0b2d0b5d0bbd0b8d0bad0b8d0b9 d0b8 d0bcd0bed0b3d183d187d187d0b8d0b9 d0b2 d180d0b0d0b7d0b4d0b5d0bbd0b5 d0b2d18bd0b1d0bed180d0b0 d18fd0b7d18bd0bad0b0, d182d0bed0b3d0b4d0b0 d18f d0bed182d0b0d0b1d180d0b0d0b7d0b8d0bb d181d0bad180d18bd182d18bd0b5 d184d0b0d0b9d0bbd18b d187d182d0bed0b1 d183d0b1d0b5d0b4d0b8d182d181d18f d187d182d0be d18fd0b7d18bd0bad0bed0b2d0bed0b9 d0bfd0b0d0bad0b5d182 d0b4d0bed0bbd0b6d0b5d0bd d0bbd0b8d0b6d0b0d182d18c d0b8d0bcd0b5d0bdd0bdd0be d0b2 d0bad0b0d182d0b0d0bbd0bed0b3d0b5 d181 d0bfd180d0bed0b3d0b0d0b9, d0b0 d0bdd0b5 d0bdd0b5 d0b2 d0b4d0b8d0b2d0b5d0bad182d0b8d0b2d0b5 language d0b8 d0b1d18bd0bb d0bfd180d0b8d18fd182d0bdd0be d183d0b4d0b5d0b2d0bbd191d0bd d182d0bed0bcd183 d187d182d0be d0bfd180d0bed0b3d180d0b0d0bcd0bcd0b0 d0bdd0b5 d0b8d181d0bfd0bed0bbd18cd0b7d183d0b5d182 d0bdd0b8 d0bad0b0d0bad0b8d185 d0b4d0bed0bfd0b0d0bbd0bdd0b8d182d0b5d0bbd18cd0bdd18bd185 d0b1d0b8d0b1d0bbd0b8d182d0b5d0ba d0b8 d0bdd0b8 d0bad0b0d0bad0b8d185 d184d0b0d0b9d0bbd0bed0b2 d0bad180d0bed0bcd0b5 EXE d182d0b0d0bc d0bdd0b5d182, d0bfd0bed181d0bbd0b5 d187d0b5d0b3d0be d0bfd0b5d180d0b5d0bcd0b8d181d182d0b8d0bb d0bfd180d0bed0b3d180d0b0d0bcd0bcd183 d0b8 d184d0b0d0b9d0bb d180d183d181d0b8d184d0b8d0bad0b0d186d0b8d0b8 d0bdd0b0 d180d0bed0b1d0bed182d187d0b8d0b9 d181d182d0bed0bb, d0b3d0b4d0b5 d0bfd180d0bed0b3d180d0b0d0bcd0bcd0b0 d0bfd0bed181d0bbd0b5 d0b7d0b0d0bfd183d181d0bad0b0 d181d180d0b0d0b7d183d0b6d0b5 d0b5d0b3d0be d0bfd0bed0b3d0bbd0b0d182d0b8d0bbd0b0, d0b0 d182d0b0d0ba d0bad0b0d0ba d0b2 d0bdd0b0d181d182d180d0bed0b9d0bad0b0d185 d0b1d18bd0bbd0be d0b2d18bd181d182d0b0d0b2d0bbd0b5d0bdd0be sysetm default d182d0be d0bed0bdd0b0 d0b1d0b5d0b7 d0bfd0b5d180d0b5d0bdd0b0d181d182d180d0bed0b9d0bad0b8 d181d180d0b0d0b7d183d0b6d0b5 d0b7d0b0d0b3d0bed0b2d0bed180d0b8d0bbd0b0 d181d0be d0bcd0bdd0bed0b9 d0bfd0be d180d183d181d181d0bad0b8 (+ d0bfd0bed18fd0b2d0b8d0bbd0bed181d18c d0b5d189d191 54 d18fd0b7d18bd0bad0b0 d0bfd0bed0bcd0b8d0bcd0be d0b0d0bdd0b3d0bbd0b8d0b9d181d0bad0bed0b3d0be), d180d0b0d0b7d183d0bcd0b5d0b5d182d181d18f d0bfd0bed181d0bbd0b5 d18dd182d0bed0b3d0be d0bdd0b5 d0b7d0b0d0b1d183d0b4d182d0b5 d0b2d18bd0b9d182d0b8 d0b8d0b7 d0bfd180d0bed0b3d180d0b0d0bcd0bcd18b d0b8 d0bfd0b5d180d0b5d0bcd0b5d181d182d0b8d182d18c d0b5d191 d0bed0b1d180d0b0d182d0bdd0be d0b2 d0b5d191 d0b8d181d185d0bed0b4d0bdd0bed0b5 d0bcd0b5d181d182d0be =) d091d18bd0bb d0bbd0b8 d18dd182d0bed182 d0bed182d0b2d0b5d182 d0bfd0bed0bbd0b5d0b7d0bdd18bd0bc?
September 11, 2013 at 5:21 pm
Anonymous
I was about to ask you what exactly were you smoking when you wrote this post. I’ll refrain given your comment above is now quite sober and makes a lot of sense.
September 11, 2013 at 5:59 pm
Kenan Kalaycı (@Kalayci_Kenan)
You are totally right that pregnant women’s preferences are ignored in doctors’ recommendations and Emily insert’s her preferences which might not represent an average woman’s preferences. It might however represent an average reader of her book’s preferences, which is not as bad.
On the other hand, it is not clear whether to ignore pregnant women’s preferences altogether is better than making an incorrect assumption about them. A lot of pregnancy advice is aimed to reduce certain risks but taken as a whole these restrictions create an enormous amount of stress on many pregnant women. And that same stress is what pregnant women are supposed to avoid!
Another missing point in Jeff’s argument is the disregard about doctor’s preferences. Jeff claims that an expert’s job is to advise on the risks of an activity but most often expert’s preferences are not aligned with her job description. If the doctor advises that it is OK to drink wine while pregnant and the pregnant woman miscarries, the doctor might be sued. Hence, it is safer for the doctor to act risk (loss, to be more precise) averse and advise accordingly. Such advise would completely disregard the benefits of an activity (drinking wine, eating sushi) and only minimise the risks.
September 11, 2013 at 8:38 pm
jeff
All excellent comments, thank you.
September 12, 2013 at 8:03 am
Stuart Buck
I have no doubt that Emily Oster is, in fact, better at interpreting statistical evidence than just about any obstetrician.
November 22, 2013 at 12:34 pm
carrot chinos
Hey! I’m at work browsing your blog from my new iphone 4!
Just wanted to say I love reading through your blog and look forward to all your posts!
Keep up the excellent work!
March 12, 2014 at 7:15 am
vegetarian recipes
I do trust all the ideas you’ve offered on your post. They’re really convincing and will certainly work.
Still, the posts are very short for novices. May just you please lengthen them a bit from subsequent time?
Thank you for the post.