Matt Yglesias and Jamelle Bouie attribute this to the hypothesis that minorities simply need a working knowledge of the majority as a "survival skill":
All that said, let me speculate a bit. To me, it’s no surprise that the highest scorers — after controlling for everything — were religious minorities: atheists, agnostics, Jews and Mormons. As a matter of simple survival, minorities tend to know more about the dominant group than vice versa. To use a familiar example, blacks — and especially those with middle-class lives — tend to know a lot about whites, by virtue of the fact that they couldn’t succeed otherwise; the professional world is dominated by middle-class whites, and to move upward, African Americans must understand their mores and norms. By contrast, whites don’t need to know much about African Americans, and so they don’t.
Likewise, religious minorities — while not under much threat of persecution — are well-served by a working knowledge of religion, for similar reasons; the United States is culturally Christian, and for religious minorities, getting along means understanding those reference points. That those religious minorities can also answer questions about other religious traditions is a sign of broader religious education that isn’t necessary when you’re in the majority. Put another way, there’s a strong chance that religious privilege explains the difference in knowledge between Christians and everyone else.
Ilya Somin isn't sure. He says that, were this the case, we'd expect the primary advantage of these groups over Christians to be regarding questions about Christianity. But while we do, in fact, hold a slight advantage in that category, the area we really clean up in is questions about "world religions" (Islam, Buddhism, Hinduism), etc.. He says this is indicative that it is great "cosmopolitanism" amongst these religious groups that accounts for the difference.
But I think Somin's test is ill-conceived. Aside from the fact that I can think of at least a few reasons why knowledge of certain "world religions" would fall in the "survival skill" category for Jews, Atheists, and Mormons, the baseline expectation would be that each group should do well in their own category and poorly in all the others. For the Bouie/Yglesias hypothesis to hold, Jews don't have to outperform Christians on Christianity -- they only need to be close to them so as to demonstrate they've attained a working knowledge of the group. The point isn't that Jews know more about Christianity than Christians (though apparently we do), it's that Jews know more about Christianity than Christians know about Jews.
Admittedly, this is somewhat difficult to test, because a lot of the questions are overlapping of Judaism and Christianity -- the only question I recall that is specific to Jews asks when our Sabbath starts (Friday, Saturday, or Sunday). It would be interesting to see if Christians did as well on that questions as Jews did on specifically-Christian questions (e.g., who founded the Protestant Reformation or what Catholic views on transubstantiation are).
But in any event, I think the results still, on face, bear out the hypothesis rather well. When you're a small, vulnerable minority, you simply have to be curious about the world around you. When you're the biggest fish in the pond, you don't. Cosmopolitanism is a survival skill.