Wednesday, October 7, 2009

More on Calorie Labeling

Megan McArdle responds to the NY calorie labeling study:
There was never any very good evidence that labelling was going to work. Most of the arguments in support seemed to rely either on self reported data, or a gut check by a handful of already pretty slender bloggers--they were sure they'd pay attention to the calorie counts, and so why wouldn't everyone else? But personal hypotheticals are at best weak evidence, and self-report is even worse. This study found that a significant minority of people reported changing their behavior as a result of the calorie information, and ordering a lower-calorie meal. But when you looked at what they actually ordered, it was no less fattening than either logitudinal or latitudinal controls.

I can think of a number of reasons for this. People may have mentally credited themselves with a savings on one item, and allowed themselves an indulgence in another: "I orderd a single instead of a double or triple, so I get large fries and a frosty!" They might just be bad at math. Or they might have wanted to look good for the interviewer, which is always a risk in these sorts of surveys. But the receipts don't lie.

There are a bunch of caveats: the study focused on poor people in fast food restaurants (on the grounds that these are the people we most want to reach.) It happened when the calorie labeling was very new, and people may have needed time to get adjusted, learning how to read the calorie counts, and remembering to do it. Public health studies of this sort are notoriously shaky, just because it's basically impossible to do a good double-blind controlled study.