It’s entirely possible that the New York Times poll out this morning is an outlier, that subsequent polls will show the traditional gender gap returning, and so on. Don’t break out the party hats when you see a poll result you like, and don’t order the hemlock when you see a poll result you don’t like.
Having said all that, the argument from the Obama campaign is that the entire poll is erroneous, and that its results indicate nothing.
Obama’s deputy campaign manager, Stephanie Cutter, drew the short straw this morning and had to argue that the poll’s finding that 67 percent think Obama announced his personal support of gay marriage for “political reasons” and only 24 percent think he did it because “he thinks it’s right” cannot be trusted.
If you want to argue that the divide is a bit closer, fine, but… here’s Cutter:
Chuck Todd: “This is such a resounding number, it’s within any margin of error you want to create… That’s a lot of people saying he did this for politics.”
Cutter: “We can’t put the methodology of that poll aside, because the methodology was significantly biased–”
Todd: “You think this is so flawed, that this number–”
Cutter: “This is a biased sample.”
Todd: “This three to one margin is somehow going to shrink down the other way?”
Cutter: “I don’t want to go through methodology on your show. I think your readers – I think your viewers would be pretty bored by it.”
Todd: “They’re junkies. They like this stuff.”
Cutter: “They sampled a biased sample, so they re-biased the same sample. I think that the results of that poll are probably pretty flawed.”
The argument is that because the Times went back to 562 of the 852 registered voter respondents they reached in April, it somehow doesn’t accurately represent the views of the electorate the way the preceding poll did. It’s possible that the 562 lean further to the right than the preceding sample. But because the Times provides the partisan breakdown of both samples, we know that the sample is essentially the same in partisan composition. It shifted from 26 percent Republican, 34 percent Democrat, and 33 percent Independent (D+8) to 27 percent Republican, 35 percent Democrat, and 34 percent Independent (D+8).
The sample of 562 is a bit smaller than one would like to see in a national poll, but it’s not wildly smaller than other national polls, and as the Times writes, “In theory, in 19 cases out of 20, results based on such samples of all adults will differ by no more than 4 percentage points in either direction from what would have been obtained by seeking to interview all American adults.”
The Times doesn’t break down its newer, smaller sample by race or age, so some might argue that the sample has too few African-Americans or young people. But for what it’s worth, the Times says they’ve taken that into account already: “Overall results have been weighted to adjust for variation in the sample relating to geographic area, sex, race, Hispanic origin, age, education, marital status and number of adults in the household. Respondents in the landline sample were also weighted to take into account the number of telephone lines at their residence. This poll also included weights based on respondents’ party identification and their presidential vote preference from the earlier survey.”
In other words, like all polls, this one could be out of whack… but there’s nothing obviously wrong with the sample.