A plurality of adults overall are also in favor, although it’s a narrow split at 45/41. Say, didn’t we see a poll just yesterday with way different numbers? Why, we did. That was NBC/WSJ, which found 57 percent opposed to the idea of a travel ban and only a thin plurality of Republicans, 42/36, in favor. How can YouGov’s numbers be so different from those?

Note that that question doesn’t mention Trump so your views of him wouldn’t color how you answer unless you already knew that he had proposed this exact policy. Anyway, on Wednesday, a third pollster (Bloomberg) found numbers for Republicans that were similar to YouGov’s — 65/22 in favor — but the overall numbers were different, with just 37 percent backing Trump’s plan versus 50 percent opposed. How do we make sense of all that? Which numbers are right?

One key difference between the polls, I think, is that the one from NBC/WSJ was conducted by phone while the ones from YouGov and Bloomberg were done online. One theory floating around to explain why Trump seems to do so much better in online polls than in ones conducted by phone is that some respondents secretly favor him but feel reluctant to admit that to a real live person. If you’ve ever heard of the “Bradley effect” (which probably doesn’t exist), you know how this goes. That was named after Tom Bradley, who led in the polls for governor of California in 1982 but ended up losing in a surprise to Republican George Deukmejian. Bradley was black, Deukmejian white. The theory afterward was that some chunk of voters were telling pollsters they were for Bradley when they were really for Deukmejian because they didn’t want the pollster to think they were racist. Once they got in the voting booth, they voted for the guy they actually preferred. You might be seeing a sort of “opposite Bradley effect” here. Trump’s been taking an unholy beating from the media for his Muslim travel ban all week. Someone who ends up on the phone with a pollster, having absorbed that media coverage, might hesitate when asked if they support it for fear of being judged — even if they really do. That may be how you end up with just 42 percent of Republicans in NBC’s phone poll backing him when the more anonymous online polls show much higher support. If that’s true then the online polls with the higher numbers of Republicans are more accurate. (And the same goes for online polls of the GOP race, which also typically show Trump doing better than phone polls do.)

That’s one way to explain this. Another way: Compare how YouGov worded the question above to the way NBC worded theirs.

YouGov’s question notes that the ban would be temporary (“until our country’s representatives can figure out what is going on”) but NBC doesn’t. Essentially, NBC asked about a permanent ban that Trump isn’t actually proposing. Like YouGov, Bloomberg also noted that the ban would be temporary — and like YouGov, they found higher levels of support. Go figure that some people would balk at a rule banning Muslims from the U.S. indefinitely but might consider one with an end date in mind.

One more bit of data, again from YouGov. They asked people how many Muslims worldwide do they think sympathize with ISIS.

Obvious differences there between Republicans and Democrats, but neither party has even 40 percent who say “very few” or “none.” In fact, there are almost as many Democrats who won’t go so far as to say “very few” (37 percent) as there are who will (38 percent). And if you look at the numbers for blacks and Latinos, two key constituencies for Dems, you’ll find that their numbers are much lower — just 21 percent of blacks and 26 percent of Latinos say “very few” or “none.” It’s white Democrats, obviously, who are pushing the overall Democratic support for those estimates way up.

In other poll news today, Trump leads Ted Cruz by, er, 27 points in Georgia and he leads by 23 nationally. According to Reuters, which conducted the latter poll, there was no change in his support after he floated his plan for a Muslim travel ban.