I would have expected breathless overinterpretation of numbers from Politico, but I’m disappointed to see Nate Silver follow suit. Once again, this shows that a little statistics and not much contextual knowledge can really send an analyst the wrong way.
The context is what, on other grounds, is a nice story by Molly Ball, comparing the partisan breakdown of early voting returns in a number of states. The problem is what Ball does with the numbers – and her error is compounded by Silver.
In both cases, they compare early voting turnout (note: more than one week before election day – more on that later) to the overall registration figures in the state (Ball, Silver) and the party ID breakdown in the state (Silver). This is then used as evidence of an enthusiasm gap.
You’d think they’d have learned. A week ago, I chided Michael McDonald for pointing to “tantalizing” early voting numbers that showed just the opposite: no enthusiasm “gap” at all. At the time, I noted that interpreting early voting numbers in a small number of states, and based on a small percentage of the eventual voters, was fraught with peril.
Now the national media have gotten in on the game. A few important things to note about early voting turnout numbers at this stage in the game:
1) There are two ways to cast an early ballot: by mail (no-excuse absentee); and in person. The problem is that Ball and Silver lump the two together, even though we have long known that GOP voters tend (for reasons not well understood) to cast far more of their early votes by mail.
2) Among the states for which Ball is able to get data, virtually all are states where most early votes are no-excuse absentee votes (CA, CO, IA, LA, NJ, NV, OR, PA).
3) If history is any guide well under 50% of the “early” votes will be cast at this point, and in many cases we have examined in the past, less than 1/4 of the early ballots will be cast one week before the election.
In short, both Ball and Silver have “discovered” that Republicans mail in absentee ballots. Congratulations! Talk to me again next Friday and Saturday, when early in-person balloting has really opened, and we have more than 10% of the ballots cast.
I have been tracking early vote returns with the team at the Early Voting Information Center. A recent exchange with Matt Damschroder, the deputy director of the Board of Elections in Franklin County, OH, unearthed a fascinating administrative experiment underway in Ohio.
Damschroder, and his counterpart in Cuyahoga County, OH, have implemented some procedures to make it easier for their registered voters to return their ballots. Both counties:
- Send no-excuse absentee applications to all voters on the NVRA list
- Paid for postage for registered voters to return their voted absentee ballot
- Cuyahoga only also is paying postage for registered voters to mail in their absentee ballot application.
Given this administrative innovation, it should be possible to examine the responsiveness of citizens to administrative outreach with respect to no-excuse absentee voting. The patterns are fascinating: self-identified partisans are dramatically more responsive to these outreach efforts.
What does this all mean? It seems to me that there is a potential unintended consequence of this administrative change – it could result in a more partisan electorate in off-year races.
We already know from previous work that no-excuse absentee ballots increase turnout, especially in low profile contests (this effect is most pronounced for permanent no-excuse absentee voting). The only remaining administrative barriers to casting a no-excuse absentee ballot are a) actually filling out the paperwork (or online form) to request no-excuse status and b) finding the stamps to mail back the request and the ballot.
Franklin and Cuyahoga have reduced the barriers just about as low as they can go. What we won’t know until the election is over is whether they unintentionally increased turnout among partisans because these citizens are most responsive to a “push” no-excuse absentee system, or whether these changes simply resort the voters (putting more partisans into the no-excuse pile and more non-affiliated into the election day pile). And, of course, we won’t know how this may propagate into other contests until the NEXT election, not something election officials want to think about right now!
To date, there are differences between D’s and R’s, but a noticeably lower return rate among non-affiliated voters, when ballot returns are calculated as a percentage of ballots requested.
Overall, 29.19% of absentee ballots requested have been returned; 30.98% among registered Democrats, 31.67% among registered Republicans, and 20.81% among non affiliated voters. The spreadsheet is attached to this post; it was created Thursday, October 14 at 4pm PT.
Nate Silver has a post over at 538 about Washington state, and the difficulties with forecasting election results there. I don’t know how pollsters adjust their likely voter models in states with significant early voting (though this seems like an increasingly important question), but I do want to note his comments about the impact of vote-by-mail. Silver contends:
Also — probably because of mail balloting — turnout in Washington and Oregon has generally been very high, so targets that might work well in other states could fail there.
This seems wrongheaded on two fronts. First, as we’ve shown several times, vote-by-mail has caused a very small turnout increase in Oregon and Washington – on the order of a few percent in federal elections. (Although admittedly, we do see a more substantial effect in lower-tier races.) In fact, political participation is part of the culture in these two states, and they have long had high rates of voter turnout.
Second, as turnout increases in a state, the impact of likely voter models presumably becomes less important. Washington and Oregon have levels of turnout way above the national average, most other states, and the pollsters’ baselines for likely voter models.
Update: Gronke posted this reply on the Time blog:
I have to chime in here. Oren is correct about Washington. The big push for vote by mail in Washington State occurred after the Rossi/Gregoire contest (2004). Fewer than 10 counties were using VBM in 2004, so any “house effects” prior to 2006 cannot be attributed to VBM.
Second, many scholars have studied the turnout effects of VBM in federal contests, and the impact is minimal, between 2-4%. Put simply, Oregon (and Washington) were high turnout states prior to the adoption of VBM and continue to be high turnout states. I’m disappointed to see Nate propagate the myth that VBM is a magic bullet for turnout. It’s a common misconception, but one that most election officials now realize is not true.
Finally, while I understand how high turnout and VBM can make the likely voter filter problematic, I don’t understand how it would be predicted to create highly variable results. A just doesn’t link to B in this blog post.
If turnout were high, then isn’t the likely voter filter irrelevant? And if high turnout plus heavy use of non precinct place voting leads to variable polling results, then Nevada, New Mexico, and Colorado should also be displaying these patterns.
Michael does a great service to all of us by posting these data, and by providing the best turnout and voter registration data available. However, my concern with these early results is that reporters won’t realize–as I’m sure Michael does–that these early votes constitute a tiny part of the overall returns, and even of the early votes.
In North Carolina, for example, only .3% of ballots have been cast (using 2006 as our baseline). In Maine, 2.3% of ballots have been cast.
Our analysis of many states in 2006 and 2008 shows that typically less than half of the early votes (not total votes–early votes) are cast more than seven days prior to the election.
Given that we already know that Republicans tend to use no-excuse absentee voting at a much higher rate anyway, it’s dangerous to conclude anything from these very early figures. I’d wait at least until Monday the 18th or better, Monday the 25th, before writing the inevitable early vote story.
A losing candidate in Everett, MA raised “the specter of voting irregularities” due to an abnormally high number of absentee ballots in the primary for State Senate.
Interestingly, the candidate, Tim Flaherty, wants the clerk to examine whether all absentee voters proved that their excuses were valid.
Ned Foley of the Moritz School of Law at the Ohio State University weighs in on our recent discussions about Alaska. For those who don’t know, Ned is one of a few experts on the legal and political wrangling in Coleman vs. Franken, and I appreciate his willingness to muse on this issue. Crossposted at electionupdates.caltech.edu.
I haven’t looked into the details of the Alaska law on this–at least not yet–and let’s hope no situation arises where the outcome matters on how a dispute over write-in ballots is resolved. But what worries me most in what I have seen as a result of the previous Election Update posts is that apparent discrepancy between the text of the statute and the view of Alaska’s Elections Director, as reported by Slate and the Anchorage Daily News.
The statute says that, to be conducted, a write-in ballot must meet this requirement regarding the write-in candidate’s name: “the name, as it appears on the write-in declaration of candidacy, of the candidate or the last name of the candidate is written in the space provided.” This statutory language is odd. It appears to permit only the candidate’s “last name” to suffice, and to disqualify any mistakes regarding a candidate’s first name. For example, “Mark Murkowki” or “Jane Murkowski” would be rejected because neither is consistent with Lisa Murkowski.
But what if the ballot has “Lis Murkowski” with the “a” missing from Lisa? Or “Liza Murkowski,” with a “z” instead of an “s”? That’s not “as [the name] appears on the write-in declaration of candidacy.” Sure, despite the discrepancy, we all know the voter’s intent in this situation. But Alaska’s statute has an important extra provision: “The rules set out in this section are mandatory and there are no exceptions to them. A ballot may not be counted unless marked in compliance with these rules.” This provision would seem to prohibit judges and election officials from giving leeway to voters in counting ballots based on discerning evident voter intent despite minor deviations from the rules.
(By the way, on the specific issue raised by Paul in response to Thad, it would seem that this statutory provision indicates that a ballot should NOT be counted if the oval is completely unmarked even if the name is correct. The statute says that write-in ballot is counted if both “the oval is filled in” and the name is correctly included. While there may are different ways to mark the oval validly, as Paul observes, not marking it at all would seem to fall outside of statutory compliance, and the provision says that no ballot may be counted unless it complies.)
Of course, this “mandatory compliance” provision doesn’t completely resolve all statutory doubt: maybe writing “Liza Murkowski”, despite the mistaken “z” for “s” is close enough to count as being the same as the name “as it appears on the write-in declaration of candidacy”. The argument is that Lisa and Liza really are the same name, at least compared to Mark or Jane. If so, then the voter would have complied with the relevant mandatory rule, and there would be no need to make an exception. But the problem with this sort of argument is that this key provision seems to tell judges and officials to err on the side enforcing the counting rule strictly, which would mean that “Liz” and “Liza” really aren’t the same as “Lisa” (despite obvious voter intent).
But the Alaska Elections Director was quoted as saying that “Lisa M” would be acceptable even if “Lisa” alone wouldn’t be enough. That position seems inconsistent with the statute. “M” is not the same as the candidate’s last name, which is required.
What I fear is litigation based on an argument that voters relied on the public pronouncements of the state’s Election Director, which appeared in the leading newspaper in the state’s largest city. Even if the Elections Director is flatly wrong under the statute, a court might accept a “due process” argument based on voter reliance. There is some case law around the country to support that kind of argument. (I haven’t looked at Alaska or Ninth Circuit precedents specifically on this point.) That kind of reliance argument, by the way, did not play a role in Coleman v. Franken, because there was no claim that voters were relying on the possibility that election officials would engage in an excessively lenient interpretation of the rules for submitting absentee ballots in Minnesota.
More recent news stories from Alaska indicate some effort to clarify the situation: http://www.adn.com/2010/09/24/1470302/murkowski-seeks-clarification.html
Let’s hope that the issue is sufficiently clarified that it does not become a practical problem. Or that the election is not close enough for the issue to make a difference.
Paul Gronke, Professor of Political Science at Reed College and Directory, Early Voting Information Center, will moderate the candidate debate between Tom Hughes and Bob Stacey for the Metro president.
Debate is sponsored by the City Club of Portland.