Early voting data: beware any conclusions
Democrats are looking at early voting data from several key states and suggesting that Election Day might not be too bad for them. But experts say the data are unclear.
So they’re taking an early peek, thanks to the advent of early voting in many states. Election officials don’t actually start tallying the votes until Election Day, but they can tell us how many people have already voted and, in states that register voters by party, their partisan breakdown. The problem is, there are so many ways to slice and dice the numbers, it’s possible to show just about anything.
But for Democrats, fighting hard against strong evidence that they will do badly in the Nov. 2 midterms, any glimmers of hope in early voting are worth a shout, if only to keep their side from getting discouraged and staying home altogether. On Monday, the Democratic Senate Campaign Committee (DSCC) put out a memo touting numbers in a handful of crucial states that – surprise, surprise – purport to show the Democrats competitive or even doing well.
"Despite national momentum being on the Republican side for months, we are not seeing anything resembling a Republican surge,” writes Sen. Robert Menendez (D) of New Jersey, the DSCC chairman. “In fact, to the contrary, in key Senate races, we are seeing encouraging signs for Democrats.”
In Senator Menendez’s breakdown, Democrats so far have cast more ballots than Republicans in West Virginia, California, and Nevada. Based on the committee’s modeling, likely Democratic voters have cast more ballots than likely Republican voters in Illinois, Wisconsin, and Washington.
All six states feature Senate races, most of them close – all of them featuring seats currently held by Democrats. The outcomes could determine which party controls the Senate come January (though for Republicans to take over, they would likely need to win five or all six of the races).
- In Nevada, the DSCC reports, Democrats and Republicans who voted for the first time in 2008 are tied in early voting so far – 6.2 percent of each group – and “sporadic” Democratic voters are slightly ahead of sporadic Republican voters, at least as a percentage of those registered.
- In Illinois, more Democrats (155,046) have voted than Republicans (55,675) so far.
- Ditto in California, where 603,713 Democrats have voted versus 545,321 Republicans.
- Also ditto in Wisconsin, though no numbers are provided.
- In Washington, which votes by mail, the DSCC’s model projects that more likely Democratic voters are voting than likely Republican voters.
- And in West Virginia, 55.5 percent of ballots cast have come from Democrats, versus 34.4 ballots cast from Republicans.
Now for some caveats: These are selective data from just a few states. And just because a registered Democrat has voted, it doesn’t mean she voted for a Democrat. Many voters never bother to change their registration, even if they philosophically left a party long ago. The DSCC also does not provide the overall partisan breakdown for voters in those states.
Nonpartisan analysts of early voting data are duking it out online over how best to crunch all the numbers at their disposal. Both Politico and FiveThirtyEight blog's Nate Silver found in weekend analyses that the early voting numbers support the polls that show a pro-Republican “enthusiasm gap.” But Michael McDonald, an associate professor at George Mason University in Fairfax, Va., wrote Monday on Huffington Post that Mr. Silver’s analysis is “fatally flawed.”
Mr. McDonald argues that it does not make sense to compare early voting totals to party registration figures. The better comparison is between current early voting numbers and those from previous years. Mr. Silver fights back, arguing that comparisons with the past could suggest “anything from an impending catastrophe for the Democrats, to a scenario in which they’d radically outperform expectations” – depending on which past elections are used for comparison.
In short, Silver concludes, early voting is so new in most parts of the country, that it’s still not possible to look at the data and really understand what’s going on.
He writes: “The notion ... that you can just plug in the 2006 early data, or the 2008 or 2004 numbers, and have some decent baseline to ‘read’ what this year’s early voting numbers mean is highly dubious.”