But the killer is that, while the study reports that it happened to liberals and conservatives alike, it also reveals that the question meant to go against the liberal prejudice (and thus that they would get wrong by answering against the data) used correct numbers presented incorrectly. That is, they showed crime rates to be higher in cities that banned handguns, whereas the actual data said the opposite.

Well, guess what, some of us know about the real world. Some of us have read reports and studies. And liberals are known to be more exposed to contravening data than conservatives (well, no wonder, when most of the media are conservative), so there is less echo chamber effect in any case.

So, I can see two possible explanations for that which do not fit Klein's argument: either the person, knowing well that the crime rate was much less where handguns were banned, simply mentally switched the proportions, or, knowing of so many other studies, reckoned that it was a trick case of not enough data to conclude against previous evidence.

I am well aware that partisan effects exist, and that they do for all. But if you feel the need to present factually wrong data on a well-known issue to test one of the groups, you cannot claim symmetry in the conclusions. Subjects highlighted where liberals tend to have views at odds with most of the published studies (most frequently, nuclear energy and GMOs) tend to be much more multi-dimensional and have much of the data published by actors that have a clear interest in the conclusions. They typically are not denial of a straight statistic of uncontroversial data. And indeed, Paul Krugman's post was moslty about the lack of simmetry in the issue, the fact that the overwhelming rejection of something that should not even be disputed is something that almost only comes from the right these days.

On a side note, Klein makes a false statement. He shows that maths skills fails to improve the chances of getting it right when the problem goes against your beliefs, but that being good at maths makes you 45% (as opposed to 25% for weak maths skills) more likely to get it right when it fits your ideology. He correctly states that maths skills increase the gap between various partisans, but then incorrectly states that being good at maths makes you more likely to get it wrong. No, not from this test at any rate. The increased gap is purely from being more likely (when good at maths) to get it right when it fits your ideology against an undifferentiated performance when it does not. Don't blame maths.