But all this changed in a version of the experiment in which the treatment was switched from boring skin cream to incendiary gun control (a law banning citizens from carrying concealed handguns in public), and the outcome was switched from rashes to crime rates. Now the highly numerate respondents diverged from each other according to their politics. When the data suggested that the gun-control measure lowered crime, all the liberal numerates spotted it, and most of the conservative numerates missed it—they did a bit better than the conservative innumerates, but were still wrong more often
...more