Mitt Romney Did Not Win A CNN Southern Whites-Only Post Debate Poll

» comments

One of several post-debate polls touting a resounding victory for Republican presidential candidate Mitt Romney has become a viral outrage on Twitter. A CNN/ORC poll showing a 67%-25% win for Romney over President Obama has come under fire thanks to a tweet, by Über-Tweeter Keith Olbermann, wondering why the poll's full results appear to show a polling sample composed entirely of Southern white people over age fifty. According to CNN Washington Bureau Chief Sam Feist, the poll did include a diverse sample, and the confusion is due to a misreading of the poll's cross-tabs.

The poll drew red flags when folks like Olbermann saw that the results in the cross-tabs of the poll's full results featured a 'not applicable' notation for every category but Southern Whites above age 50. He tweeted:

Can @CNN explain this? RT @PlainRane CNN snap poll comprised entirely older, southern whites. (page 11)

Maybe there's a good reason for 'N/A' (not available or applicable) in every non-South, non-White breakdown in @CNN' http://tl.gd/jhdbam

As it turns out, there is a good reason. CNN Washington Bureau Chief Sam Feist tells Mediaite, via email, that 'our poll is actually very very strong in terms of demographic diversity and breakdown for a post-debate poll. TPM got a lot of folks talking about it because they misinterpreted a cross-tab chart.'

'The N/A in the demographic columns were there because the sample size was not sufficient for cross-tab demographic comparisons. Every reputable pollster uses N/A in its crosstab columns like that in order to make sure that nobody tries to make scientifically valid comparisons using relatively small sample sizes. But TPM mistakenly interpreted the data to suggest that we had not polled in certain demographic or geographic groups.'

In other words, some of the demographic data wasn't sufficient to draw conclusions from, but whatever spreadsheet they use for cross-tabs has to put them in some column, and put something in the other columns. TPM has updated their post on the subject, Feist says, and 'blessed' their poll. As he did for TPM, CNN also provided Mediaite with the demographic breakdown of the polling sample, which does, in fact, include people other than Southern whites, although the sample of non-whites was significantly smaller than the proportion of non-whites in the U.S. population, but consistent with other post-debate polls of debate watchers that have been done over the years. This is likely a reflection of the makeup of the debate audience.

Feist added that 'our poll was solid and well crafted. ORC pre-selected its sample before the debate, insuring an appropriately diverse sample.'

A CBS News post-debate poll showed 46% giving Romney the win, with 22% calling the debate for the President, and 32% calling it a tie.