I wonder what would happen if they unpacked 'liberal' and 'conservative' into their component parts and repolled.

I hope (but not necessarily expect) that a lot of the discrimination would come out as social scientists being 'biased' against specific positions whose advocates claim to be conservative, that make no sense from a social science perspective - and not against other conservative positions with no such implication. These could be swept up by bandwagon formation.

It might possible to find certain other disciplines where the participants would have a similar bias against liberals due to some particularly popular boneheaded positions claimed by their advocates to be liberal, and again bandwagon formation.

Or it could just be a liberal dominance of social science, complete with discrimination. A more detailed poll might help to elucidate this.