[A powerful extension of the LV critique, plus an explanation of why the RV numbers are widely wrong as well: certifiable Republican oversampling. It's not just the CBS poll. Gallup's got the same problem. And although it seems like a million polls came out this last week, 95% have just been those two, quoted in a million different ways.]
Public Opinion Watch - Sept. 22, 2004 Ruy Teixeira
Decoding the Gallup and New York Times Polls
Gallup poll of 1,022 adults, released September 17, 2004 (conducted September 13-15, 2004)
CBS News/New York Times poll of 1,287 adults, released September 18, 2004 (conducted September 12-16, 2004)
-----------------
Here are Bush's leads in the three national polls released before Gallup's current poll (no registered voter [RV] data available for Democracy Corps and Harris, only likely voter [LV] data; Pew and Harris matchups include Nader):
Democracy Corps, September 12-14 LVs: +1
Pew Research Center, September 11-14 RVs: tied
Harris Interactive: September 9-13 LVs: -1
Looks like a tie ball game, right? But according to the Gallup poll conducted September 13-15 and released September 17, Bush is up . . . by thirteen points???
Let's just say I'm just a wee bit skeptical of this one. First, Gallup's poll only includes one day (September 15) that the three other polls do not cover, so it can't be Gallup's survey dates that explain the big Bush lead.
Second, this thirteen-point lead is an LV figure and, as I've repeatedly emphasized, Gallup's LV screening procedure produces completely untrustworthy measures of voter sentiment this far in advance of the election. Here is a summary of the case against Gallup's LV data:
Sampling likely voters is a technique Gallup developed to measure voter sentiment on the eve of an election and predict the outcome, not to track voter sentiment weeks and months before the actual election. There is simply no evidence, and no good reason to believe, that it works well for the latter purpose. In fact, the evidence and compelling arguments are on the other side: that the registered voters are the more reliable gauge of voter sentiment during the course of the campaign.
Here's why. Gallup decides who likely voters are based on seven questions about their interest in voting, attention to the campaign, and knowledge about how to vote (e.g., where their polling place is located). The interested/attentive/knowledgeable voters are designated "likely" and the rest are thrown out of the sample. But as a campaign progresses, the level of interest among voters tends to change, particularly among those with partisan inclinations whose interest level will rise when their party seems to be mobilized and doing well and fall when it is not. Because of this, partisans of the mobilized party (lately, Republicans) tend to be screened into the likely voter sample and partisans of the demobilized party (lately, Democrats) tend to get screened out. But tomorrow, of course, the Democrats could surge, in which case their partisans may be the ones over-represented in likely voter samples.
That suggests the uncomfortable possibility that observed changes in the sentiments of "likely voters" represent not actual changes in voter sentiment, but rather changes in the composition of likely voter samples as political enthusiasm waxes and wanes among the different parties' supporters. And that is exactly what political scientists Robert Erikson, Costas Panagopoulos, and Christopher Wlezien find in their analysis of Gallup's 2000 RV/LV data in their forthcoming paper, "Likely (and Unlikely) Voters and the Assessment of Campaign Dynamics" in Public Opinion Quarterly: "shifts in voter classification as likely or unlikely account for more observed change in the preferences of likely voters than do actual changes in voters' candidate preferences."
That means that, instead of giving you a better picture of voter sentiment and how it is changing than conventional registered voter data, likely voter data give you a worse one since true changes in voter sentiment are swamped by changes in who is classified as a likely voter.
I think the case against the Gallup LV data looks rock solid. In my view, it's time for them to drop reporting these data because they are highly likely to give an inaccurate picture of the state of the race and, by doing so-especially given the high profile of Gallup's polls-unfairly pump up one side of the race and demoralize the other. That doesn't seem acceptable to me.
Of course they'll reply: well, our data work so well right before the election, they must be the best data to use all the time. But, for the reasons outlined above, that reasoning is completely specious. And then there's this: the LV data haven't been working so well lately even right before the actual election. In three of the past four presidential elections (including the most recent one), Gallup's final RV reading was actually closer to the final result than their final LV reading!
As I have repeatedly argued, it's time for a serious rethink down at Gallup headquarters.
Throwing out the Gallup LV data, then, let's move on to their RV result: an eight-point Bush lead. Obviously pretty far off the results of the other contemporaneous polls summarized above, but . . . could be, I suppose.
But then there's this: the Gallup internals show Kerry with a seven-point lead among independent RVs. Huh? Kerry's losing by eight points overall, yet leading among independents by seven. How is that possible? Only if there are substantially more Republicans than Democrats in the sample.
That suggests that reweighting the sample to reflect the 2000 exit poll distribution (39 percent Democrats/35 percent Republicans/26 percent independents) would give a different result. It does: the race then becomes dead-even, instead of an eight-point Bush lead. (Note: Steve Soto of The Left Coaster got Gallup to give him their party identification distributions for this poll and confirms a five-point Republican party identification advantage in their RV sample.)
One other note: I mentioned the Pew Research Center poll had the race dead-even just as in the reweighted Gallup data. And what was Pew's party identification distribution in their RV sample? You guessed it: a four-point lead (37 percent to 33 percent) for the Democrats, just as in in the 2000 exits.
Right after the Gallup poll was released, the latest CBS News/New York Times (CBS/NYT) poll came out. That poll, conducted September 12-16, also gives Bush an eight-point lead (50 percnet to 42 percent) among RVs, but also, not coincidentally, gives the Republicans a four-point edge on party identification. Reweight their data to conform to an underlying Democratic four-point edge, as in the 2000 exit poll, and you get a nearly even race, 47 percent Bush/46 percent Kerry.
Nearly even. That goes along with the 46 percent to 46 percent tie in the Pew Research Center poll (which gave the Democrats a four-point edge on party identification without weighting) and the 48 percent to 48 percent tie in the Gallup poll (once weighted to reflect an underlying Democratic four-point edge). Not to mention the two other recent national polls (Harris, Democracy Corps) that show the race within one point.
Perhaps all this is just a coincidence, but the pattern seems striking. Once you adjust for the apparent over-representation of Republican identifiers in some samples, the polls all seem to be saying the same thing: the race is a tie or very close to it.
Looking past the horse race result, there is another aspect of the new CBS/NYT poll that deserves emphasis. As Chris Bowers of MyDD points out in his insightful new essay, "Rapid Poll Movement Is a General Election Myth," (http://www.mydd.com/story/2004/9/18/454/00913#readmore) the new CBS/NYT poll is actually a lot worse for Bush than the CBS News poll released just a week earlier. That's because, since the current poll is substantially more Republican than the earlier poll (which actually had a slight Democratic edge), Bush should actually have performed better than the earlier poll on the horse race and on indicators like job approval and right/direction wrong track in this poll, instead of about the same. That also means that if we adjust the current poll to correct the apparent surplus of Republicans, Bush's performance on these indicators should actually decline below the measurements of the earlier poll.
Since CBS News thoughtfully provides the overall result and the result broken down by party identification for each and every question in their survey, it is possible to estimate what Bush's ratings would have looked like if there weren't so many Republicans in the sample. Here are some examples, based on reweighting the current poll to the 2000 exit poll distribution of partisanship:
Overall job approval: 49 percent approval/44 percent disapproval Economic job approval: 42 percent/52 percent Iraq job approval: 45 percent/51 percent Campaign against terrorism job approval: 57 percent/37 percent Right direction/wrong track: 40 percent/53 percent
In every case, these ratings are worse than they were a week ago, making the idea that the race is tightening up more plausible.
Of course, Kerry needs not just a tight race, but to pull ahead. Given Bush's continued vulnerabilities, which these data highlight, Kerry's got the opening to do so. We shall see if he is able to take advantage of this opening.
====
The Century Foundation is nonprofit and nonpartisan and was founded in 1919 by Edward A. Filene. [Filene, a Jewish Boston Brahmin who was one of the most prominent figures during the golden age of American aristocratic liberalism around the turn of the century, is credited with one of my favorite quotes. Inventor of the bargain basement, he reportedly said "Why shouldn't the American people take half my money from me? I took all of it from them!"]