(This article was originally published at Statistical Modeling, Causal Inference, and Social Science, and syndicated at StatsBlogs.)
Alex Gamma writes in with what he describes as “an amusing little story” from two years ago:
When Deaton & Case published their study, and after your re-analysis had uncovered the missing age-correction, I’ve pointed out this issue to several news blogs that reported on the study, but were not aware of the problem (only about 2 in 5 or so responded).
One was the blog of Swiss National Television, titled “Many American men die in their prime”. They brought an interview with their U.S. correspondent about the increasingly dire living condition of these white men and how that might drive them into suicide. I pointed the problem out to them, linking to your Slate piece and adding a relevant quote.
The reply I got was (my translation):
The study was conducted by a Nobel prize winner in economics, published in PNAS [link to paper] and discussed in pretty much all the serious media, e.g. by Paul Krugman in the New York Times.
That’s also described in the Slate article. It’s true that there has now emerged a discussion about the interpretation of the numbers and the right or wrong numbers, respectively. The interview was simply about the content of the study, how it is to be understood and how it is discussed in the USA.
So far, that’s the not all too unexpected, dodging of the issue. But what they added at the end was almost brilliant:
In the last paragraph of the Slate article itself it says fittingly: I’m pretty sure some people will find major problems in my analysis. That’s how it goes, two steps forward and, if we’re lucky, only one step back.
They beat you with your own weapon!
At least that’s what I thought back then. I thought “don’t put a disclaimer that you don’t really mean, it might backfire”. Of course, I don’t know how you meant that disclaimer, probably it was dead serious.
Hmm, good point. I write “I’m pretty sure some people will find major problems in my analysis,” but that was too strong. Here’s what I should’ve said:
I wouldn’t be surprised if people find major problems in my analysis. And in any case I’m sure our understanding will evolve as we learn more. That’s how it goes, two steps forward and, if we’re lucky, only one step back.
Hmmm . . . “Our understanding will evolve as we learn more,” that’s a bit too vague. But it’s an improvement on my original phrasing. People do sometimes find major problems in my work, but saying “I’m sure” that would happen, yeah, that’s not accurate. Most of the time I think my published claims hold up.
The post Two steps forward, one step back appeared first on Statistical Modeling, Causal Inference, and Social Science.
Please comment on the article here: Statistical Modeling, Causal Inference, and Social Science
