Interesting, and if anything, probably understating the effect -- most obviously, gun control doesn't directly overlap with conservative/liberal political self-identification, and the specific matter of concealed carry is even less directly correlated when your average age is nearly 50.
On the other hand, the lack of interviews to see how people explain their decision-making process is... not ideal. Scientific procedure is good and all, but I'd be very surprised if at least some of the folk with strong opinions on this matter did not read the question as what they'd suggest instead of what the study itself says, and that brings us more to questions about whether their pre-existing beliefs are well-formed. I know I'd find it very hard to point to the 'wrong' answer just because one new study contradicts thirty older ones, especially a new study without citations (and, since gun control usually focuses on /violent/ crime instead of general crime, not the most useful one).
The graph clustering is probably an even more disheartening part of the story. The "rash cream" version of the question is not a very difficult math problem, and nearly sixty percent of the survey population got it wrong, disturbingly close to simple random selection. Compared to that, the politically-motivated innumeracy is almost a drop in the bucket.
gun control doesn't directly overlap with conservative/liberal political self-identification
???
Original article: Motivated Numeracy and Enlightened Self-Government.
Easy-to-read summary of the results: Science Confirms: Politics Wrecks Your Ability to Do Math.
Easier-to-read summary of the summary: A group including both liberals and conservatives were first tested for numeracy, then given either fictional stats on the efficacy of a rash treatment (how many people get better/worse, with/without the treatment) or quantitatively identical fictional stats on gun regulation and crime (in how many cities crime went up/down, with/without gun control laws). They had to say whether, according to the stats, the treatment was effective or whether the laws reduced crime. (In half of the cases the answer was yes, in half no). For the rash problem, numerate people did better in giving the right answer, irrespectively of their politics. For the gun control problem, numerate people (of both political sides) were much better at giving the right answer when it agreed with their political beliefs; when the conclusion implied by the stats disagreed with their political beliefs, they did no better than innumerate people.