DANIEL SAREWITZ devotes an entire column in Slate to the fact that only 6% of scientists self-identify as Republicans (55% self-identify as Democrats) without offering even a hypothetical explanation of how this situation might have come about. This is as close as he gets:

I can think of three testable hypotheses they might look into. The first is that scientists are hostile towards Republicans, which scares young Republicans away from careers in science. The second is that Republicans are hostile towards science, and don't want to go into careers in science. The third is that young people who go into the sciences tend to end up becoming Democrats, due to factors inherent in the practice of science or to peer-group identification with other scientists. In the absence of data, I leave it to you to decide which you find most plausible. But by all means, social scientists should look into this.