Polling Variation

Polls were recently conducted regarding Trumps proposal to temporarily stop immigration of Muslims. Response was initially generally favorable outside of the New York Times building. A second poll phrased the question differently and got a strong negative reaction. The difference was in the phrasing of the question asked as I found out today from WSJ columnist J. Taranto:

What accounts for the divergence between the two polls’ results on the Muslim-exclusion proposal? Here is the question the WSJ/NBC poll posed: “Recently, Donald Trump has called for a total and complete shutdown for any Muslim being allowed to enter the United States. Do you favor or oppose this proposal or do you not have an opinion one way or the other?”
And here’s how CBS asked it: “Do you think the U.S. should temporarily ban Muslims from other countries from entering the United States, or not?”
These are both accurate descriptions of the Trump proposal, but the emphasis is quite different: “total and complete” vs. “temporarily.” The WSJ/NBC pollsters also mentioned Trump by name, making it likely that respondents’ attitudes toward him weighed more heavily in the results. And the WSJ/NBC question included the no-opinion option, whereas CBS primed its respondents to form an opinion.

An honest pollster will be able to tell you what people think. A good pollster will tell you what you want to hear to +/-3% accuracy.