I have been regularly following the polls on the Real Clear Politics site. And I have been noting that Mitt Romney, though still behind on average, was pulling even with Barack Obama. A couple of days ago, the spread was less than 1%. Then, suddenly, the gap opened up a bit - to 2.3%. Given that this is an average of the nine most recent polls, even that small a change is big, so I looked at the details. It seems that one poll, by Bloomberg News, came in, in which Obama leads by 13 percentage points, 53% to 40%! Without that Bloomberg poll, the difference between Romney and Obama is still under 1 point.
This poll seems to have overrepresented Democrats. The Democratic favorable-vs.-unfavorable figures in this poll were 50-43; the Republican, 41-50. I am not certain what it is nationwide, in fact, but these numbers are suspicious. And in the favorable-vs.-unfavorable ratings in the Bloomberg poll, Obama is rated favorable 55-42, and Romney unfavorable 39-48. This is clearly an outlier. In an average of 7 polls (including this one) RCP shows Obama rated favorable 52-44, Romney unfavorable by such a small amount that to the nearest percentage point it's a 43-43 tie! (If you exclude the Bloomberg poll and look at the other six, Obama's favorable-vs.-unfavorable figures are 51-44, Romney's 43.5-42, actually positive!)
Why is this poll such an outlier? I wonder. It doesn't seem to be Bloomberg as such; I've looked at some earlier Bloomberg polls and they do not seem as far out of line. (I've noted that Rasmussen's polls, generally, tend to skew a bit toward the Republicans.) This is merely an enigmatic outlier.
This poll seems to have overrepresented Democrats. The Democratic favorable-vs.-unfavorable figures in this poll were 50-43; the Republican, 41-50. I am not certain what it is nationwide, in fact, but these numbers are suspicious. And in the favorable-vs.-unfavorable ratings in the Bloomberg poll, Obama is rated favorable 55-42, and Romney unfavorable 39-48. This is clearly an outlier. In an average of 7 polls (including this one) RCP shows Obama rated favorable 52-44, Romney unfavorable by such a small amount that to the nearest percentage point it's a 43-43 tie! (If you exclude the Bloomberg poll and look at the other six, Obama's favorable-vs.-unfavorable figures are 51-44, Romney's 43.5-42, actually positive!)
Why is this poll such an outlier? I wonder. It doesn't seem to be Bloomberg as such; I've looked at some earlier Bloomberg polls and they do not seem as far out of line. (I've noted that Rasmussen's polls, generally, tend to skew a bit toward the Republicans.) This is merely an enigmatic outlier.
No comments:
Post a Comment