http://www.nationaljournal.com/blogs/hotlineoncall/2013/06/gallup-post-mortem-leads-to-polling-changes-04
Nobody not named Romney lost bigger last November than the Gallup Poll. They got their ass kicked by the results of the election, which they missed on badly - bigger than any other reputable poll. They have now published the result of a self-study that identifies 4 problems that contributed to their bad performance: 1) how they weighted likely voters, 2) underrepresentation of the East and West coasts in geographical controls, 3) underrepresentation of nonwhite voters, and 4) issues in how it contacted landlines that resulted in an “older and more Republican” survey sample. The problem of course is that most of these problems were pointed out to them last fall, and they were arrogant in dismissing them. The two biggest problems, which are related, was in their estimate of likely voters and weighting of non-white voters. It may or may not have been coincidence, but this is basically the same problem that most Republican professional political operatives had: Their conventional wisdom was that the 2008 voter turnout was a fluke, a result of one-time voter fatigue with 2 wars and a cratering economy. They could not believe that Black, Hispanic and other first-time voters would turnout in anything like 2008 numbers in 2012. Others (including, but not limited to, Nate-the-Great Silver) cited longer term voting and demographic trends in assuming that ethnic and new-voters would turn out, and they were right. Some of this is just reasonable disagreement over polling methodology, but a significant portion of it is clearly based on assumptions about who "real" Americans are. Do this date, many Republicans believe that Obama is an illegitimate president (quite aside from any judgement about how effective he has been) because he did not get the support of a majority of *real* (read for the most part white, over 40, non-coastal) Americans. Gallup clearly played into these assumptions with their polling model, perhaps because at some level the organization shared them. Some might think this story deserves a "No TV" label, but I don't think so. As many of the stories about this today have pointed out (and as Silver has been explaining for some time now) the main reason for these kinds of polling inaccuracies is that the incentives in political polling are not really to be accurate, but to get mentioned in, and legitimate, television news pieces. The 2012 presidential election was not a landslide, but it was never really very close, and there was remarkably little movement in the actual numbers from mid-summer through election day. But that does not make for very interesting news stories, so television is incentivized to give more weight and focus to any poll that shows movement or differs from the pack, even though by definition those are just the polls that are most likely to be wrong. The cable networks, which all saw nice rating increases during the election season, are unlikely to care much about how wrong they were, since by now almost nobody who watches them really expects them to be accurate. Thus they have little reason to change. Gallup on the other hand has a business model that goes far beyond simply predicting the presidential horse race every 4 years. Most of what Gallup does to make money is predict things that do not have such a clear criterion of accuracy. When they fall on their ass so glaringly in the presidential prediction game, in undermines customer confidence in the accuracy of the other products that are their bread and butter. There is a chance that Gallup will make changes to be more accurate next time - and if they do, it is likely that we will not see as many news stories about the Gallup poll (or aggregations that are disproportionately influenced by them), and instead will see news stories driven by more interesting and polarizing and (probably) inaccurate polls. -- -- TV or Not TV .... The Smartest (TV) People! You received this message because you are subscribed to the Google Groups "TV or Not TV" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/tvornottv?hl=en --- You received this message because you are subscribed to the Google Groups "TVorNotTV" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/groups/opt_out.
