Testing hypotheses

(Hat tip: The Volokh Conspiracy.)

Tutissima Cassis looks at the widely-circulated news that Republicans are more likely than democrats to believe certain falsehoods.

A number of left-leaning bloggers have pointed to surveys that show that Bush supporters are more likely to believe erroneous information about Iraq (such as whether weapons of mass destruction have been found) than are Kerry supporters. The surveys appear to be real. Bloggers such as Brian Leiter and Philocrites point these out. I'm disturbed by the implication that pundits and bloggers are all too ready to make -- namely, that Republican voters are, as a group, less well-informed than Democratic voters. Indeed, it seems like this survey may be unconsciously designed to inadvertently come up with that kind of result. Why? Well, we know that most of the electorate simply doesn't follow the issues. Within each party, there is some percentage of people who simply don't know what's going on in the world. That is, there are Democrats who know political facts and Democrats who don't, and there are Republicans who know political facts and Republicans who don't. This survey asks exactly the kinds of questions that the group of Democrats-who-don't-know are likely to answer correctly, but Republicans-who-don't-know are likely to get wrong.

Now, how to construct a valid test?

Even assuming an equal distribution of ignorance, if Republicans are asked questions that the least well-informed members of their group are more likely to get wrong than the least well-informed Democrats, the survey will give results that appear to indicate that Republicans are less well-informed. The results of the survey can and are interpreted to suggest that Republicans as a group are less well-informed than Democrats, when in fact the much more innocuous explanation is a distinct possibility. It should be relatively easy to test whether this is in fact the cause of the survey results. All that is necessary is to turn around and ask questions that the least-well-informed Democrats are probably more likely to answer wrong than the least-well-informed Republicans. (Again, using the simple tool of "I like Bush" or "I don't like Bush" to determine how the uninformed are likely to respond). For example, "Is it true that filibustered Republican judicial nominees Miguel Estrada and Priscilla Owen received a rating of "not qualified" from the American Bar Association?" (It is not true. Both received unanimous ratings of well-qualified). Or perhaps, "Is it true that President Bush has done nothing to try to stop the genocide in Sudan?" (As Kristof points out, while Bush could have done more, he has so far done more than any other American president in this area, and has been a global leader in trying to stop the killing in Sudan.)