Once in a while, PPP’s sample will end up less Democratic than the 2008 breakdown; its Wisconsin sample splits 33 percent Democrat, 32 percent Republican, and 34 percent independent; on Election Day 2008, this state split 39 percent Democrat, 33 percent Republican, and 29 percent independent.
In all these cases, the PPP sample includes proportionally fewer independents than the most recent election. Independents are traditionally less active in midterm elections; on the other hand, most pollsters have found the number of Americans who self-identify as independents increasing since 2008.
Will Democrats really make up a larger share of the electorate in 2010 than they did in 2008, when the Obama campaign poured enormous resources into its get-out-the-vote effort, and droves of Democrats and first-time voters came out to elect the first African-American president?
Jensen told NRO that comparing exit polls to standard pre-election telephone polls “is complete apples and oranges”; he says he never uses them as a guide for his polls. “Yes, our [Kentucky] poll had 52 percent Democrats and the 2008 exit poll showed only 47 percent Democrats,” he said. “But only 59 percent of the Democrats in our poll yesterday say they voted for Obama, whereas the exit poll showed 69 percent of Democrats voting for Obama. . . . If I call you up today and ask you who you’re voting for, you might call yourself a Democrat, but if I ask you to fill something out right after you voted for John McCain you’re less likely to call yourself a Democrat.” That is indeed possible, but it is worth noting that other pollsters find fewer Americans identifying themselves as Democrats since 2008, not more, and more voters identifying themselves as independents.
The partisan breakdown of each state’s and district’s electorates will shift from year to year, so pollsters are entitled to some leeway in their samples. But at some point, a poll sample just tilts too heavily to one party to be a reliable guide to the electorate as a whole. A poll of New York voters that had a sample in which half the respondents identified themselves as Republicans would be widely dismissed, and deservedly so; there just aren’t that many GOP voters in the state. None of the samples from PPP are quite that outlandish, but they tend to presume much lower turnout among independents (plausible, considering the lower level of interest in non-presidential years), similar turnout among Republicans (unlikely, considering the indicators of high GOP enthusiasm), and higher turnout among Democrats (hard to imagine, barring some unexpected new factor that rejuvenates Democratic enthusiasm to its 2008 peak).