A new paper by David G. Blanchflower, Andrew J. Oswald, and Sarah Stewart-Brown argues that consuming more fruit and vegetables increases psychological well-being. This is the kind of paper that usually sends me on a rant about over-interpreting results and not recognizing that ceteris non paribus. But these guys get it exactly right – here are the last two sentences of their abstract:
Reverse causality and problems of confounding remain possible. We discuss the strengths and weaknesses of our analysis, how government policy-makers might wish to react to it, and what kinds of further research — especially randomized trials — would be valuable.
This is indeed an interesting, suggestive result – it’s not something I would have thought to look for – and I think the topic merits more study. Kudos to the authors for realizing the limitations of their work.
I want to point out one particular kind of reverse causality that doesn’t appear to be mentioned in the paper: reporting biases. Other studies have found that when you have people self-report behaviors, people who are happy/healthy/otherwise doing tend to over-report their “good behavior”. This is particularly true for retrospective reporting. Reporting biases of this kind helped create the famous result that fiber intake reduces the risk of colorectal cancer. More careful research has found no link whatsoever.
Another thing that comes to mind is that I recall hearing about a couple of programs in the US that try to encourage people to consume more fruits and vegetables, including at least one which was randomized. These would be great sources of variation for further research on the healthy eating-well being connection.
Hat tip: Chris Blattman