Don’t Make Me Say I Told You So: It’s a Bump, Not a Surge

By Stuart Rothenberg,

The ink on my Sept. 18 column warning against overreaction to new polling and alleged “surges” had not yet dried when USA Today released its newest Gallup Poll, predictably asserting the existence of an alleged Republican surge. Others quickly joined the chorus.

First, let’s stipulate that recent polling has shown some improvement in Republican numbers, whether it’s in the president’s standing or the generic ballot. President Bush’s job approval numbers bottomed out earlier this year and have been inching up for a couple of months.

That’s the good news for Republicans, and any good news is better than bad news. But if you look at presidential job approval and the generic ballot, it is awfully difficult — no, actually, it is impossible — to conclude a dramatic surge has occured or is under way. That doesn’t mean it won’t happen, only that it hasn’t occurred so far.

The USA Today/Gallup poll found Bush’s job ratings among all adults at 44 percent approve/51 percent disapprove. That’s an improvement from an Aug. 18-20 USA Today/Gallup survey that showed the president’s ratings at 42 percent approve/54 percent disapprove. So, Bush’s approval is up 2 points and his disapproval is down 3. Clear enough.

A Bloomberg/Los Angeles Times poll showed the same thing, and it’s a development worth noting. But remember, Bush’s disapproval was still at 51 percent in the USA Today poll and 52 percent in the Bloomberg/L.A. Times survey, and the improvement in Bush’s numbers is still modest. It’s not as if the public has suddenly changed its evaluation of the president. A majority of Americans still disapprove of his performance, and that’s not a good place to be for the Republicans going into a second midterm election.

In evaluating presidential job approval ratings, Gallup actually reports two additional sets of numbers, one based on three-poll rolling averages and a “smoothed approval” estimate, based on work done with a consultant, Yale professor Don Green.

Both reports show a slow but steady increase in Bush’s job approval from a stretch in May, when it bottomed out in the low 30 percent range, to now, when it is around 40 percent. They show no surge, however, and Gallup Poll Editor in Chief Frank Newport wisely notes that after years of doing polling, he pays less attention to the results of a single survey and more to trends and patterns.

Now we get to an even bigger problem for those who want to see a surge in the USA Today/Gallup numbers — the generic ballot.

Far too much has been made of the USA Today/Gallup poll’s finding that the generic Congressional ballot is tied at 48 percent among “likely voters.” Unfortunately, some observers have compared that number to previous Gallup generic ballot numbers, proclaiming incontrovertible evidence of a surge.

However, Newport told me recently that the Sept. 15-17 USA Today/Gallup survey is the first one this year that screened for “likely voters.” Previously, Gallup reported either on registered voters or “regular voters.” Regular voters and likely voters are not identical, Newport said, since different questions are used to identify both subsamples.

So, if you must compare apples to oranges, go right ahead and compare “regular voters” to “likely voters.” Just remember that it’s not kosher.

If you look at Gallup’s registered voters going back almost three months, you will find that the current 51 percent Democratic/42 percent Republican generic ballot in mid-September is almost identical with Gallup results from Aug. 10, July 30 and July 9. So the only “surge” among registered voters appeared in a dramatic narrowing of the generic in a single Aug. 18-20 poll. But that single survey is an outlier, and the USA Today/Gallup generic hasn’t really moved in months.

Interestingly, less than 48 hours after the USA Today/Gallup poll became public, a CBS News/New York Times survey was released. While it reported on the opinions of adults, rather than registered or likely voters (who usually are more Republican), the CBS News/New York Times survey found the president’s job approval at 37 percent, “virtually unchanged” from August. Moreover, the poll found Democrats with a 15-point advantage in the generic ballot, 50 percent to 35 percent.

Does all of this mean terrible news for Republicans, since they have been hoping for a surge? Not necessarily.

It’s odd that there hasn’t been more comment about the difference between the USA Today/Gallup registered voters and likely voters generic ballot results. While the poll showed Democrats with a 9-point advantage among registered voters, the generic ballot was even among likely voters.

With all of the talk about Democratic enthusiasm and Republican division, I would not have thought there would be such a large gap between the two samples.

Elections are won and lost by actual voters, not by hypothetical voters. So, if Democrats don’t beat the GOP on turnout on Election Day, the generic ballot among all adults won’t matter at all.


This column first appeared in Roll Call on September 25, 2006 Copyright 2006 © Roll Call Inc. All rights reserved. Reprinted with permission.