I cited the "6% of scientists are Republicans" figure in various discussions and articles a few years ago. At the time, I also noted that there was some uncertainty as to how accurate the figure was, given that it was based on a sampling of AAAS members. That concern has grown in recent months, precisely because I lately have become an AAAS member in order to subscribe to the journal Science. I already knew that becoming a member does not require being a scientist (I am a part-time science journalist) but the distinction between "AAAS member" and "scientist" has been underscored for me by (a) actually attaining membership for the price of a magazine subscription and (b) receiving subsequent mailings inviting me to further professional memberships, such as in a chemical society, and even being told I was "specially selected" as an AAAS member or some such.
Lest I be misunderstood, I continue to believe that the 6% figure, though clearly not a precise measurement, points to a real problem of heightened alienation between scientists and Republicans. But seeing it repeated now, years later, with no qualifications, does not much incline me toward Krugman's argument, to wit that there's no sign of academia moving left, it's just that crazy Republicans moved right. The current GOP overall is indeed too right-wing for me, as I've made clear, but craziness at one end of the political spectrum often begets and is abetted by craziness at the opposite terminus, and pretending that's not so is a sign you may be succumbing to it yourself.
BTW I came across Krugman's piece via this tweet, with which I sympathize greatly, by Jonathan Haidt:
I just discovered I'm an "outraged conservative," according to Krugman. I thought I was a calm, civil centrist. https://t.co/8ta0lSdXUC— Jonathan Haidt (@JonHaidt) January 4, 2016