Turns out that the participation the huge US military survey on DADT is self-selected:
Sunday was the deadline for troops to complete the Defense Department's "don't ask, don't tell" attitudes survey, and officials at the Pentagon said the final tally on completed responses was 109,883 -- a response rate of only about 27.5 percent.
A guy I know in the Air Force got the survey and he filled it out (he told me his responses... they weren't half bad). But if I'd hazard a guess the people who filled it out were those who felt strongly on the topic, and we all know that the people who feel strongly on this topic tend to be on the opposite side of the spectrum from the rest of us.
We don't really know who were the 27.5% were, but when you have a response rate that low (of people who knew in advance what the topic of the survey was), it does introduce a new element. Why would some people choose to respond and others wouldn't? Who knows, but whatever the reasons were, they may affect the end result of the survey.
The Stars and Stripes article linked above does mention who's better at filling out these surveys: