A close examination of Obama's recent good polling

President Obama’s approval rating and, more specifically, the difference between his disapproval and approval rating (the “spread”) has seen a dramatic improvement in the past few days. Over at RealClearPolitics, we see the spread has gone from 7.8 to 11.7 in just two days. In fact, the change in his approval rating of 0.85/day (53.2 vs 51.5) is the best positive rate of change for President Obama since late February. (As an aside, at several times, his approval rating has dived at a faster rate, e.g. mid-August, mid-July, see RCP chart).

What could be causing this recent “good news”? First, we can look at the expanded list of polls here to see which polls are currently being used to calculate the approval and disapproval ratings and the spread. At the time of filing, RCP is using the following:

Gallup 8/31 – 9/2 1547 A 55 39 +16
Rasmussen Reports 8/31 – 9/2 1500 LV 47 53 -6
CNN/Opinion Research 8/28 – 8/31 1010 A 53 45 +8
Ipsos-McClatchy 8/27 – 8/31 903 RV 56 40 +16
CBS News 8/27 – 8/31 1097 A 56 35 +21
Pew Research 8/20 – 8/27 2003 A 52 37 +15

Two polls stick out: Rasmussen Reports and CBS News. The approval and disapproval numbers for the former poll are both around 1.75 standard deviations away from mean; it’s clearly an outlier. But, in that regard, so is the disapproval number from CBS News (one full standard deviation off of mean): if not an outlier, it’s by far the second most deviant poll result behind Rasmussen. Furthermore, as the CBS News approval rating does not differ from the Ipsos-McClatchy value (56), this results in a similarly-suspect spread (+21) that is likely too high.

Why are these two polls off so much? There are a couple reasons why Rasmussen typically polls low for the President. First, to be honest, Rasmussen may be truly biased toward conservatives. But assuming a faithful polling, it could be that Rasmussen typically polls likely voters (LV) as opposed to registered voters (RV) or simply adults (A).  In fact, if we compare Obama’s approval rating found by Rasmussen and Democracy Corps up to late June (end of DC data), we see little difference. However when these two polls are compared to Gallup and CBS News (both A-type polling), we see the LV-polls are generally 5 points lower than A-polls over the past six months. More specifically, the correlation between Gallup and Rasmussen is 0.75, meaning the two polls are in strong agreement about the changes in Obama’s approval ratings, but they disagree on the magnitude of his rating at any point.

What explains the problem with the CBS News poll? To reiterate, the approval rating found by CBS News and Ipsos is exactly the same; where CBS differs is in too low a disapproval. Put another way, CBS News is allowing more people to not commit to the two choices: approve or disapprove. This highlights a second way Rasmussen Reports stands out. If we add the number of “approve” and “disapprove” for each pollster, we see that all polls average 91.7 (with a standard deviation of 3.87); thus, 8.3 percent of those polled say “other”, “don’t know” or “neither”. CBS News, though, averages 87.8 approve+disapprove — once again, a full standard deviation off mean. We can conclude CBS News is allowing more respondents to non-commit. How does Rasmussen stack up? Over the year, Rasmussen is averaging 98.7 approve+disapprove, two standard deviations above mean. Also, they are the only poll to commit 100% of respondents to one of the two choices at any time this year. We can conclude then that Rasmussen is doing a better job getting an opinion from those polled.

Back to our original question: what explains the recent upturn in the spread? Two reasons: 1) The inclusion of low disapproval ratings from CBS News and Gallup, both of which are likely cases of under-reporting and 2) the substitution of the Gallup approval rating (55) for NBC News (51). Will the recent trend hold? If history — or at least recent history since February — is any indicator, no. But, we’ll find out soon enough if these two days are the exception or if the President’s approval ratings have finally bottomed out. In the meantime, we’ve learned two things: one, Rasmussen is not only polling low for the President, it’s also polling “tougher” than others while CBS News is simply substandard and two, we’re looking at a two-day trend in Obama’s approval rating not seen in six months. Clearly times and polling have been rough for the President.