We Were Looking at the Wrong End of That Berkeley Study
We assumed they were measuring politics right and peronality wrong. I suspect it was the other way around.
After looking at the actual paper, I noticed they derived the “LIB/CON” from seven different measures, some of which could be fudged and none of which included the most objective measure of all: voting behavior. They also went into great detail about which personality components were correlated with the LIB/CON difference but ignored which components of LIB/CON went with which personality difference.
Conservatives who don't adhere to leftist stereotypes of conservatives might not answer questionnaires. A typical libertarian might look at a questionnaire that asks about “opposition to government policies to aid minorities,” mutter something about being asked if he stopped beating his wife, and fold the questionnaire into a paper airplane instead of answering it. (The same libertarian might be inclined to work more diligently for deregulation to aid minorities.)
John J. Ray has more authoritative comments.