Checking party IDs

+ More

The following comes from the Gallup organization:

Americans are about as likely to identify as Republicans as they are Democrats according to a review of recent Gallup polls. However, once the leanings of independents are taken into account, the Democrats gain an advantage. Democrats have been on par with, or ahead of, Republicans in party identification since the second quarter of 2005.

I'm not sure why this is treated as news. Since Gallup pioneered random-sample polling in October 1935, Democrats have always had an advantage over Republicans in party identification. When I entered the polling business in 1974, Democrats had a huge advantage in party ID, something on the order of 49 percent to 25 percent.

The real news came in 2004, when the NEP exit poll showed party identification of the electorate as 37 percent Democratic and 37 percent Republican—the best showing for Republicans since 1935. In the release referenced above, Gallup is saying that party ID is now 33 percent Democratic and 32 percent Republican but that self-identified independents tend to lean Democratic. In any long historical perspective, this represents a big gain for Republicans and a big loss for Democrats. In contrast, the difference between these numbers and party ID in the 2004 NEP exit poll is statistically insignificant. (Hat tip to Ed Morrissey,

As Morrissey notes, many recent public polls have shown much greater Democratic advantage in party identification, as much as 10 percent. Their samples are far out of line with the NEP exit poll. How to explain these differences?

One way is to say that party identification could actually have changed substantially in the 16 months since the 2004 election. Historically, party ID seldom changes so much so rapidly, but it could. Certainly you could list a series of reasons—Iraq, Katrina, Dubai ports—that could plausibly explain such a shift. But still it seems unlikely.

Another way begins with noting that most current public polls sample "all adults." Some sample "registered voters." When we get closer to election time, pollsters narrow down their samples from AAs to RVs or LVs ("likely voters"). In recent years LVs have been significantly more Republican than RVs, who, in turn, have been significantly more Republican than AAs. The NEP exit poll was somewhat more Republican than most LV polls at the time. In this view, current AA polls might well be consistent with the NEP exit poll; they just measure different groups. Party identification might not be changing at all.

In 2002 and 2004, Republicans were more likely to turn out and vote than Democrats. That's partly the result of the Bush and Republican National Committee turnout efforts, and it also probably reflects greater enthusiasm by Republicans than Democrats (hard as that is for the mainstream media to believe). But there's no guarantee that Republicans will have a turnout advantage in November 2006. On the contrary, there are some reasons for believing that the Republican base today is discouraged and less likely to turn out than in 2002 and 2004. Polls don't do a good job of projecting turnout; at best, they just provide clues.

My own bottom line: I tend to discount polls that show a big Democratic lead in party identification. But I also take them as indicating that Republicans may not have the advantage in turnout in 2006 that they had in 2002 and 2004. May not: We'll see.