If the question isn't meaningful neither is the answer

Nate Silver of the website FiveThirtyEight turns his attention to that biased survey of military personnel about the gay ban. Silver doesn't pile on to how homophobic the answers are (though he might discuss that later). Instead, he talks about how useless the questions are. Here are some of his complaints:

The survey doesn't ask if a soldier knows if his colleagues are gay, it asks whether he believes he might be gay. A soldier might answer yes because of a couple factors, neither of which have anything to do with whether the colleague actually is gay. Those factors are:

* A soldier in a unit with low morale may decide it must be that way because somebody is gay -- we've heard so much about how gay soldiers supposedly affect cohesion and morale. But it would be low morale triggering suspicion, not fact leading to low morale.

* A soldier might conclude a colleague is gay based on stereotypes. Many are teens who haven't been far from home and aren't aware of the broad range of the way people act. If you don’t act like I do you must be gay.

Again, there is a difference between knowing if a colleague is gay and believing so. This is inappropriate for the survey because those who take up that speculation are likely more homophobic than the average soldier. It is also improper because it is based on the reaction to a closeted gay colleague, and not based on one who is serving openly.

The survey reports speculations, not opinions. It doesn't even ask what soldiers think of the ban itself.

In a related article, Andrew Gelman also of FiveThirtyEight, points to research that shows how false-positives (soldiers believing colleagues are gay but actually aren't) can swamp meaningful survey statistics.

I'll let The Onion have the last say on the issue, at least for today. If you can take it.