[Posted by Karl]
I generally enjoy Charlie Cook’s work, which made his bunkum on independent voters all the more disappointing:
It’s misleading to say that the state of the economy determines whether a president will win reelection. But it is fair to say that when a White House incumbent is running for a second term, the election is first and foremost a referendum on that president; the single most important factor that voters consider in assessing a president is the state and direction of the economy. That is the default factor unless something happens to shift a race’s dynamic and make the election more like a choice than a referendum. At least, that’s what I’ve always thought.
But now I wonder whether the economy will drive this election to the usual extent—or to the extent I had thought. More specifically, will the Republican Party nominate a candidate who can credibly compete for the independent voters whose support is so important in general elections?
Independents represented 29 percent of the electorate in 2008. In last year’s combined Gallup polls, though, they were 40 percent—a record high. In 2000, Republican George W. Bush won the independent vote by 2 percentage points over Democrat Al Gore but narrowly lost the overall popular vote. In 2004, Democrat John Kerry actually carried independents by 1 point but lost the national popular vote by 3 points. The winner of the independent vote doesn’t necessarily win the general election. But a candidate has to be very competitive among independents to have a chance to win. In 2008, the GOP’s John McCain lost the independent vote by 8 percentage points and the election by 7 points.
As political scientists like Alan Abramowitz have noted, Cook’s basic thrust is incorrect and undercut by his own examples. The major fallacy here is thinking of “independents” as a cohesive voting group, when they are anything but:
Let’s start with the claim that independents make up the largest segment of the American electorate. That’s true only if you lump all independents together including those who don’t vote and those who lean toward a party. In 2008, according to the American National Election Study, independents made up 40% of eligible voters but only 33% of those who actually voted. Moreover, of that 33%, only 7% were true independents with no party preference. The other 26% were leaners.
And what about those independent leaners? Fully 87% of them voted for the candidate of the party they leaned toward: 91% of independent Democrats voted for Barack Obama while 82% of independent Republicans voted for John McCain. That 87% rate of loyalty was identical to the 87% loyalty rate of weak party identifiers and exceeded only by the 96% loyalty rate of strong party identifiers.
It’s hardly surprising that the vast majority of independent leaners voted for their party’s presidential candidate in 2008. The evidence *** shows that independent Democrats and Republicans held very different views on major issues — views that were very similar to those of their fellow partisans. Independent Democrats were more liberal than weak Democrats and about as liberal as strong Democrats while independent Republicans were less conservative than strong Republicans but just as conservative as weak Republicans.
Actual party identification is fluid. Cook’s own examples from 2000 and 2004 should have suggested to him that de facto Democrats may have felt more comfortable identifying as such at the end of the Clinton administration, but may have felt less so — and de facto Republicans more comfortable identifying as GOP — in the first post-9/11 presidential election. The same underlying dynamic likely accounts for Jimmy Carter’s narrow victory in 1976 while losing the independent vote in the first post-Watergate presidential election.
Moreover, as political John Sides has noted (again contra Cook’s premises), the relationship between the economy and elections it is stronger among independents than among partisans. Sides has scatter plots to back him up, but this is also common sense. Sides notes “[p]artisans are happy to vote for their party under most any circumstance and often rationalize their view of the economy accordingly.” I would add that other issues, e.g., national security and social issues, also may motivate partisans (including leaners) to vote party. If pure independents had fixed or passionate views on such issues, they would be partisans, not pure independents. Cook’s own 2008 example only shows that when the economy is plunging into a deep recession, independents will vote against the party holding the White House. Shocka.
That bit of sarcasm aside, I do not mean to be harsh on Cook. Political scientists would not be writing about these mistakes unless plenty of other pundits and journalists had been making them already.
These mistakes are enabled in part by pollsters, who often do not ask questions to identify leaners. Gallup does screen this way in their presidential approval polling, which currently shows pure independents giving Obama 37% job approval. That’s up about 10% from last autumn, though below Obama’s 42% approval among all “independents” and his 45% overall approval number in the Gallup poll.
Pew also screens independents for leaners and will report results for them, yet generally does not report results for the remaining unaffiliated respondents. Cook mentions a recent Pew poll in his column:
Republicans should be concerned that Mitt Romney’s numbers among independents have been tanking in recent weeks; he went from double-digit leads over Obama in some polls, including one by the Pew Research Center, to a 9-point deficit.
The Pew poll to which Cook refers reported Obama leading Romney 52%-44% overall, and 51%-42% among “independents”. However, the sample in this Pew poll was 49% Democrats + leaners, and 39% Republicans + leaners. Thus, it’s entirely possible that Romney led among pure independents in this poll. Moreover, the WSJ/NBC News poll just two weeks prior had a sample that was only 45% Dems+leaners and 37% GOP+ leaners, with perhaps twice as many pure independents, producing a 49%-44% margin for Obama. This is not to accuse the Pew poll of having an improper sample (although at this point the number of polls showing Obama over 50% are small). While we might argue about the party ID in a sample compared to registration or past election results, I am unaware of any appropriate standard for measuring leaners. Indeed, it is the apparent inability to determine an appropriate mix of independents for any given poll that makes an undue focus on “independents” a dodgy proposition.