Have people seen Andrew Kohut's Times op-ed today? He argues that the polls were wrong on Obama/Clinton because a) lower income whites are less likely to answer surveys and b), in adjusting for this long-standing fact, pollsters failed to take into account the greater racism of those low-imcome whites who refused to answer compared to those who did.
If he's right, then it would seem there was no late break for Hillary, and all the business about Hilary's showing emotion had nothing to do with the result. What do you think?
At a rough guess, "stunning impact of media event" is really "polling not that accurate" about 99.9% of the time...
ReplyDeleteKarl Rove penned a similar theme saying we Democrats are beer drinkers not wine consumers. Hey - pour me a pint!
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteLike the op-ed, some further analysis would have to be done to see exactly why the discrepancy exists, BUT I do find it highly admirable that somebody is watching and evaluating the actual worth of the polls. Too may people take numbers as gospel.
ReplyDeleteI see no reason to believe that low income whites refused to answer pollsters there more than elsewhere. I do suspect that there is some of the scond point, which has been labeled the "Wilder effect."
ReplyDeleteBut, I think it is in fact mostly women pouring out for the almost tearing up Hillary.
Barkley
There was a piece on Andrew Gelman's blog on this topic yesterday. It argued for the same conclusion (i.e. that the polls were just wrong and there was no swing), but for different reasons related to the "voter screen", and potential problems in the survey weighting. The race explanation was discounted, though not necessarily convincingly.
ReplyDeleteI, too, heard the race explanation on NPR (~compared to previous numbers Obama did well in very white NH.) as well as the warmish weather getting out far more numbers of the elder more conservative Hilary supporters.
ReplyDeleteBut nothing like depressing me with the reminder that although this was a record turnout, it was still 44% --more than half think this is a circus not worth bothering about.
Unless there is some reason to believe that low income whites in Iowa are less racist than low income whites in New Hampshire I don't find the argument compelling. I suppose there could be an income effect with low income whites more likely to vote than to caucus but would that variation in participation rates be enough to mask the racism factor in Iowa as compared to NH? No one can deny that race is a factor in this race as is gender, candidates are going to both gain and lose votes from each. But trying to identify the precise role of either in any given race seems on the whole useless.
ReplyDeleteBut if you simply insisted on asking the question of what could suppress polling participation among low income whites a pretty good answer would be 'football'. We had four NFL playoff games on the weekend and the BCS Championship on Monday. I can almost guarantee you the result you would get if you called during game time and a guy answered, click.
Of course: the pollsters over-looked the NFL roster!
ReplyDeleteVery interesting:
ReplyDeletehttp://www.slate.com/id/2181849/fr/rss/
Obama got the percentage the polls predicted for him. He had no falloff. The error was all in underestimating Clinton's vote, particularly among women.
ReplyDeleteThe rest is all bullshit.