washington, dc

The Democratic Strategist

Political Strategy for a Permanent Democratic Majority

Ruy Teixeira’s Donkey Rising

Adventures in Likely Voter Land

It has not escaped my notice that many people are puzzled as to how exactly polls go about determining “likely voters” (LVs). There’s a good reason for this: polling firms or sponsors rarely put much effort into explaining, clearly and precisely, the mechanics of how they select these LVs.
So, as a public service, here’s how they do it. Let’s start with Gallup. According to David Moore of Gallup:

Gallup asks each [RV] respondent seven LV screening questions, and gives each person an LV score of 0 to 7. [Assuming a turnout of 55 percent], the top 55% are classified as likely voters.

In practice that typically means all of the “7”s–given full weight–plus some proportion of those with lower scores (usually the “6”s), who are weighted down so that the size of the likely voter sample matches the projected turnout for the year (apparently 55 percent this year). All other voters are discarded from the sample.
What are the Gallup likely voter questions? Unfortunately, the exact questions and their wording are not released by Gallup along with their polling data, but the questions apparently involve past voting behavior, interest in the election, intention to vote in the election and knowledge of things like the location of the local polling place.
That’s how Gallup does it. What about other organizations–do they select likely voters in the same way? Nope, they don’t. CBS News doesn’t use a cut-off model, where low-scoring respondents are thrown out altogether, but instead includes everyone in their RV sample, in some form, in their LV sample. They do this by asking respondents a series of voting-related questions and then assigning each respondent a weight based on their score on these questions, from very high weights for high-scoring respondents to very low weights for low-scoring respondents.
Finally, by far the most common way is simply to ask a few screening questions and then terminate the interview with those respondents who give the “wrong” answers. Or only one question; some likely voter screens are as simple as asking an RV how likely they are to vote in the upcoming election; if they don’t say “almost certain” or “probably”, out they go.
So that’s how they get the likely voters in the polls you read about. How do they know that likely voters, months before the election, are actually the voters who will show up on election day? They don’t.
Here’s David Moore from Gallup again: “We simply do not know, nor can we know, which model is better during the campaign itself. ” Exactly. So why does he think the Gallup LV model works so well months and months before the election. Because “if it is the most accurate model just before the election, it is probably the most accurate during the campaign as well”.
But that doesn’t follow at all. The Gallup LV model could work perfectly right before the election (not that it really does, but that’s another discussion) and still be quite a biased instrument earlier in the campaign. Pretty much by definition, Gallup’s LVs months before the election are not the same voters as Gallup’s LVs right before the election, since voters answer the LV questions differently at different stages of the campaign. And if there is any kind of partisan dimension to “tune-in”, so that, say, Democratic partisans or groups that lean strongly Democratic (like minorities) tend to tune in later, that means the LV model will have a systematic tendency to, on average, favor the party (the Republicans) whose partisans or groups tune in the earliest.
Of course, my hypothesis here about Gallup LV bias might be completely wrong. But to evaluate it, Gallup would have to make available the demographics and partisan breakdown of the both its RV and LV samples for the polls it releases plus, ideally, the results (including demographics and partisan breakdowns) of the various screening questions it uses. I’m not holding my breath.


Welcome to the New Donkey!

I just wanted to urge everyone to point their browsers to the just-started “unofficial” DLC blog, NewDonkey, written by my good friend, Ed Kilgore, one of the most astute political analysts around. His blog is invariably well-written and full of valuable insight (his coverage of the Republican convention has been particularly good). So visit often. You’ll be glad you did.


So What’s Going on in Pennsylvania?

There’s been a shocking amount of weeping and wailing and gnashing of teeth in Democratic circles considering that most polls still show Kerry ahead. Indeed, the just-released ICR poll (see below) has Kerry ahead by 7 points among RVs, a greater margin than the same poll had three weeks ago.
But it’s hard to cheer people up when they’re determined to be gloomy. Indeed, there’s a tendency to seize on “evidence” that supports a gloomy viewpoint without looking at it very closely.
Take the case of Pennsylvania and recent polling data. Democrats seem to be convinced the race there has seriously tightened (see this post by Josh Marshall which is otherwise terrific and level-headed but buys into this particular meme).
What’s the evidence for this tightened Pennsylvania race? Disregarding the recent poll by the very Republican Strategic Vision firm, the most recent reputable poll in PA is the Gallup poll of 8/23-26. That poll shows Kerry trailing Bush by a point among likely voters (LVs). I emphasize that this result is among LVs because most PA polls that people have heard about have been conducted among registered voters (RVs). Therefore, comparing these earlier PA polls to Gallup’s current LV result is not an apples to apples comparison and tells you nothing about whether and how the race has changed.
Especially when we note the following: the same Gallup poll that has Bush ahead by a point among LVs in the Bush-Kerry matchup has him behind by 5 points among RVs!
In fact, check out the last three reputable poll results from PA among RVs:
8/23-26 (Gallup): Kerry, +5
8/13-21 (Issues PA): Kerry, +2
8/11-16 (Quinnipiac): Kerry, +5
Conclusion: there is no tightening in the PA race once we do an apples to apples comparison.
This is only the latest example of confusion being sown by Gallup’s LV model, which has been producing consistently pro-Bush results lately. Indeed, Gallup did a WI poll at the same time as their PA poll and found Bush ahead by 3 points among LVs in WI, but Kerry ahead by the same margin among RVs.
I continue to believe these LV results should be de-emphasized until later in the campaign. It is still too early to put much faith in likely voter screens/models as representing very accurately the voters who will actually show up on election day. There is reasonable evidence that careful likely voter methodologies work well when it is quite close to an election and do fairly accurately capture that pool of voters. But there is no such evidence for LV samples drawn this far out.
Indeed, my understanding is that Gallup, in particular, does LVs throughout the campaign not so much because they believe they are capturing election day voters all along, but more so that they can avoid having to explain sudden shifts in the horse race question as LV data replaces RV data. There have apparently been some problems with this in the past, so reporting both from the very beginning of the campaign eliminates any potential embarrassments along these lines. But that doesn’t mean the LV data is any better at this particular point in time–it merely means they’re providing it.
In fact, since the sample size for LVs is smaller and since the composition of the LV sample will shift depending on how political developments are affecting interest and intensity levels among different groups of voters, additional volatility and potential bias is built into the LV samples that is not there with the RV samples.
The issue of bias can particularly be a problem if it’s true that, say, Democratic-leaning voters tend to tune into campaigns later and are therefore more likely to be screened out of the likely voter sample until close to the election (especially with a tight screening procedure, which Gallup definitely has). If that’s true, then an LV sample could perform reasonably well close to election day, as a greater proportion of Democratic-leaning voters finally get screened in, but be quite biased toward the Republicans until then.
That could be what’s happening today.
So take a tip from me: always check those RV results. It’ll help keep a smile on your lips and song in your heart!


The Race at the Start of the Republican Convention

The Myth: The SBVT controversy seriously harmed the Kerry campaign. Bush comes into his convention in much better political shape than he has been for quite a while.
The Reality: The race has changed little since the start of the SBVT controversy. Bush enters his convention with basically the same political vulnerabilities he had previously.
Let’s go to the numbers. The poll that best provides a before-SBVT damage and after-SBVT damage picture of the horse race is the Gallup poll. That’s because Gallup polled both on August 9-11–about a week before media coverage of SBVT really heated up–and on August 23-25, right after the coverage peaked and just as the Kerry campaign began its push-back.
What do the Gallup numbers show? As Gallup’s release on their latest poll succinctly puts it: “No Change in Presidential Race Despite Attack Ads“. Just so.
In fact, to the extent their numbers show change, it’s in the opposite direction to the one everyone is assuming. In their August 9-11 poll, Kerry was behind by one point (47-48) among RVs; in their August 23-25 poll, Kerry’s ahead by a point (48-47). (Bush’s approval rating also declines by 2 ponts between the two surveys).
So why were (and are) people so convinced SBVT hurt? There were the Annenberg numbers, of course, on how many voters had heard of the the SBVT charges and found them at least a little bit believble. But there’s a lot less to these data than people assumed–see EDM’s earlier analysis on how these numbers were widely misinterpreted.
There were the August 23-25 Gallup numbers on likely voters, showing Bush ahead by 3 points, that fed the impression Bush was pulling ahead. But these LV numbers also represented no change from previous Gallup polls, which had showed Bush ahead among this group by about the same margin. (Indeed, it’s interesting to note that in the entire month of August only one poll–Gallup–showed Bush ahead among LVs in the Bush-Kerry matchup and it did so three times and by almost identical margins. Must be something going on with that Gallup LV model.)
There was the Los Angeles Times (LAT) poll, which showed Bush with a 3 point lead among RVs, released right after the peak of the mudslinging. But the LAT poll had no point of comparison in August, much less close to the beginning of the SBVT controversy, so the LAT result showed nothing about change due to SBVT. Moreover, the LAT result was an outlier among the month’s RV polls–every other poll taken during the month (save one Gallup poll)–had Kerry tied or ahead in the Kerry-Bush matchup.
Finally, there is the most plausible–in my view–source of this sentiment: the fact that a number of polls show a tightening of the horse race between very early August (i.e., right after the Democratic convention) and late August. That tightening ranges from 2-7 points, turning a small post-convention Kerry lead into a smaller Kerry lead or tie, depending on the poll you look at. But the most plausible hypothesis for this tightening is a natural post-convention decay in Kerry’s support (given a lack of new impetus in Kerry’s direction) over the course of the month, rather than the specific effect of the SBVT brouhaha.
So where does that leave us? In my view, about where we were before the Democratic convention. In fact–in addition to the horse race–if you look at Kerry-Bush comparisons on issues and on personal characteristics, the results of a number of polls seem almost to replicate the results of that particular poll prior to the Democratic convention.
And another critical thing hasn’t changed at all–Bush’s ratings in all his vulnerable areas (the economy, Iraq, health care, etc.), as well as voters’ sense of whether the country is going in the right direction and whether a different direction is needed. These indicators have all continued to be quite negative (in some cases, have actually worsened) over the course of August, including the period allegedly affected by the SBVT controversy.
This is Bush’s problem. He’s got to run on something and, unfortunately for him, he has precious little to run on other than being the president of 9/11. The SBVT ads and subsequent media feeding frenzy didn’t change that equation in the slightest–and it’s not an equation that favors Bush’s re-election.


NPR Survey Shows Kerry 50%, Bush 46%,

A Aug. 22-24 survey of registered voters conducted for NPR by Greenberg, Quinlan,Rosen and Public Opinion Strategies found John Kerry Leading George W. Bush 50% to 45% in a two man race and Kerry 47%, Bush 43% and Nader 3% in a three way match-up.


New Annenberg Center Study Shows 46% Believe Bush Behind Attack Ads, 37% Do Not

A Aug 23-26th Poll by the Annenberg Center for Public Policy shows a plurality of Americans – 46% – believe President Bush was behind the ads attacking John Kerry’s military record while only 37% believe the Bush campaign’s denials.
Day by day tracking of the percentage of voters who were influenced by the accusations and came to doubt that Kerry deserved his medals showed that from August 10-15 the percentage of doubters hovered in the low 20’s, then rose between August 16-22 (reaching almost 30% on August 18th) and then returned back down to the low 20’s between August 20-25.


Florida Poll Shows Race Too Close To Call – But Trend Appears Favorable for Kerry/Edwards

A CNN/USA Today/Gallup poll of 1002 Floridians conducted Aug. 20-22 shows registered Florida voters split evenly split 46%-46% between Kerry/Edwards and Bush/Cheney.
The trend among these voters since July 19-22, however, is toward the Democrats with support for Bush/Cheney declining from 49% to 46% while Kerry/Edwards rose one point from 45% to 46%. The same trend is evident among likely voters as well.


Zogby Poll sees Kerry Leading in Most Battleground States

An August 23rd. Zogby survey of key battleground states found Kerry in the lead in most. In some cases the lead was strong, outside the margin of error of the polls; in others only a small gap separated the two candidates.
The key results are shown below
Kerry Lead in Battleground States
Michigan – 5.2%
Pennsylvania – 8.3%
Wisconsin – 4.4%
Minnesota – 5.7%
Iowa – 7.0%
New Mexico – 5.6%
Washington – 8.4%
Oregon – 11.3%
Arkansas – 2.6%
Missouri – 0.5%
Nevada – 1.7%
Tennessee – 1.9%
Florida – 0.6%
Bush, in contrast, was ahead in only two battleground states, Ohio (51.4% to 45.8%) and West Virginia (49.3% to 41.5%).