The Residual Fallacy

Man, this is starting to get on my nerves. It's obviously wrong. It's even a special case of a fallacy well-known as such: the false dichotomy. I'll call it the residual fallacy. Here comes an example, from the famous paper "Age and the Explanation of Crime" by Hirschi and Gottfredson, published in the world's most renowned sociology journal (p. 571):
The factors Greenberg adduces to explain desistance offer a plausible account of crime rate differences between adolescence and early adulthood, between, say, 19 and 24, but they do not provide a plausible account of the similar decline in crime rate [sic] between, say, 29 and 34.
Case closed! The fallacy here is to argue that if a variable or set of variables does not explain 100% of the variance in a dependent variable (i.e., if there are residuals), then we may conclude that it explains 0%. Of course, we may not. There are a lot of numbers between 0 and 100.

(This is rather surprising from authors who correctly argue, in the same paper, that theories which distinguish between offenders and nonoffenders can be useful even if they do not simultaneously account for the age effect on crime.)

One sees this quite a bit. Watch out! Fallacy! That's my service announcement for today. You're welcome. And here's a funny link on mating strategy.

No comments: