The Election Assistance Commission’s Election Administration and Voting Survey has been released. This is the first in a series of posts that will highlight some patterns and anomalies in the data.
The EAVS is one of the best ways to assess whether or not a state is adhering to the requirements of the National Voter Registration Act of 1993, which obligates states, among other things, to provide the option to register to vote via motor vehicle agencies and other social service agencies.
To assess compliance, however, the data need to be reported. I have shown below a table that reports the state by state totals from three variables in the EAVS that should in principle have the same value:
- QA5a – “The total number of registration forms received by your jurisdiction”
- QA6_Total” – “Registration forms received, broken down by source”, the reported figure should be a sum of the individual sources, but is also labeled explicitly on the questionnaire as “QA5a”, alerting the jurisdiction that that total here should match the total listed above.
- Regtotal – My own calculated total of registration forms from all sources.
The data are reported by state below. As you can see, there are only eleven states where all three figures match as they are supposed to: AL, CO, CT, LA, ME, MI, MN, NC, NH, OR, and WY. ND is not required to report this information. These states get an “A+” for reporting.
Wisconsin simply forgot to enter the “total” for QA6_Total, but the numbers match. We’ll give them an “A”.
Idaho, New Jersey, and South Dakota reported nothing for the NVRA section at all. Not sure they can get a grade other than “F”.
I haven’t probed the other differences in order to give more nuanced grades. I’ll leave that to other experts.
It’s not often that one gets to brag about knowing an academic whose work is cited in a Supreme Court opinion. But today, our friend and colleague Peter Miller has his very own SCOTUS citation:
Further media exposure for Peter can be found on Last Best News, a site dedicated to telling the stories about the people and culture of Billings and eastern Montana. Yes, Peter is from Billings, MT.
Congratulations, Peter. We’re excited to see where else your work will prove so useful.
Hillary Clinton plans to call for an early voting period of at least 20 days in every state, the Washington Post’s Anne Gearan reports. This is not a Democrat’s first foray into the voting wars this election season, but it is Clinton’s first specific policy recommendation about election administration reform. A recent Harvard Law Review note argues that federal early voting legislation may be viable, so we do well to view her announcement today at Texas Southern University as more than just liberal hand-waving.
If Clinton wins the election and Congress passes this legislation, what might be its impact?
Sixteen states already have early voting periods that begin at least 20 days before the election. This change would likely impact these states the least.
Twenty-four states and DC have early voting periods that are shorter than 20 days. Their election codes would need to be changed and they may need to redirect resources into early voting.
Ten states have no early voting at all. A federal mandate would significantly affect these states.
Would a standard of at least 20 early voting days increase turnout?
The Clinton proposal–at least what we can glean from the pre-release for the speech–leave many details unanswered. For early voting, the devil is very much in the details.
For example, what about weekend voting, notably early voting through the final Sunday before Election Day?
Would legislation mandate a minimum number of hours per day, or across the full 20 day period?
Would there be any formula to require a minimum number of early voting locations, and how would states calculate this formula–by CVAP? geographic dispersion? Commuting patterns?
Prior research shows that these details are critical when determining how much early voting costs and whether different segments of the population use this convenience voting method.
Nonetheless, two points seem clear. First, that if this legislation passes, it would significantly change the importance of early voting across the country, since campaigns would have much greater incentives to focus on the early vote nationwide
Second and related, this would create a uniform early voting period across the United States. Greater uniformity across our hyper-federalized system is not always a good thing, but in this case, it makes it easier for voters to know when to vote and may foster cooperation between election administrators across the states
Hat tip to Rick Hasen; Derek Muller has written a really wonderful primer on the constitutional history of representation and redistricting in the United States, masquerading as a commentary on the Evenwel case.
Required reading not just for interested students and scholars, but any citizen (and should we add non-citizen? under 18? disenfranchised felon? non-registered yet eligible??) in the United States.
I’m looking forward to reading the papers and hearing about new research into election administration next week. This is an open, public conference; not sure if proceedings will be on the web. Papers should start to appear next week.
A recent local election in Red Willow County, NE is “credited for high turnout,” according to the lede (and attributed to claims by local elections officials).
Officials also praise the all-mail system that allows them to “keep the voter registration system updated, given the number of ballots returned undeliverable.”
Results were as follows:
Ballots Sent: 4735
Ballots Returned: 1821
Official Turnout: 38%
Turnout after undeliverable: 45%
2007 Special Turnout 31%
On the face of it, this seems impressive. But the devil is in the details.
It’s plausible that the voting by mail system increased turnout by 50% (from 31% to 45%) but making that calculation (presuming “the vast majority” of undeliverable ballots are voters who moved out of the county) raises issues about how the voter registration system is maintained and what election officials will do with these results.
The obvious first question is: why were 713 ballots undeliverable?
Election officials are quoted as saying that “the vast majority of the returned ballots stem from voters who relocated out of the area without updating their information with the clerk’s office.”
The implication is that 15% of the voter registration rolls in this county (or the state?) consists of deadwood–individuals who are registered in two NE counties or have moved out of state or who are deceased. 15% is a pretty high number for a rural county without a lot of in- or out-migration. Do we know if these individuals have simply moved within the county (for instance)? Do we know if these individuals have active registration records in two counties in NE?
Post election, local officials in this county may consider first, what they’ll do with this information. In fully vote-by-mail states, an undeliverable ballot prompts a forwardable postcard notifying the voter that their registration record is out of date. This follow-up step is critical.
(Officials imply that they are using the results to clean the rolls, but don’t say how.)
Second, officials should run the undeliverable records through the statewide registration system to assure themselves that voters aren’t registered in two counties.
The Montana field experiment has been deemed a campaign finance violation by the chief regulator in Montana.
I’m not sure any other outcome was expected, politically, and I’m also not sure this will stand up legally (or be pursued at all by the state attorney–they may just leave this statement of a violation as the appropriate slap on the wrist).
But as I wrote back when the controversy erupted, the bigger concern for academia is how we got here at all.
This has sparked a lot of discussion within academia, including a number of panels at MPSA and the CASBS, with more to come (including a symposium in PS if I can get my act together!). The most positive outcome may be a bit of circumspection and modesty among academic researchers.
Working my way through Electoral Studies recent releases (thanks for the RSS feed!) came across an interesting analysis by Shaun Bowler of a survey conducted by the Public Policy Institute of California (PPIC).
The survey asked a series of detailed questions about whether or not voters were happy with the amount of information that they had about a set of initiatives. But perhaps more interesting, the survey provided the opportunity for voters to explain the reasons they voted for some of the initiatives, and the results are pretty encouraging for those who argue that citizens can accumulate enough information to cast an informed ballot.
Most striking to me were two things.
First the relatively low percentage of “don’t knows” across 34 initiatives that had been on the ballot, shown below (hey Shaun, look up the “s1mono” scheme in Stata).
Second, voters were provided a list of reasons that they voted for the initiative legalizing marijuana. The reasons were, well, reasonable, and interest group information (the most oft-cited source of voter information) ranks very low on the list. (Sorry about the poor quality of the screen grab.)
All in all, a nice piece summarizing a lot of survey results over a decade, looking at voting on referenda and initiatives in California.
Full piece available here.
They have an interesting design, combining pre-election and post-election surveys and an exit poll–the former allow them to evaluate the impact of voting for the winner (or lower) on changes in perceptions of integrity.
The measures of integrity are are follows:
Our dependent variable measures citizens’ confidence in the integrity of the election. For the pre[post]-electoral survey, we use the following question: “In your opinion, how clean will [were] the presidential elections be [held last July 1st]?” Respondents chose among the following options: “Very clean,” “Somewhat clean,” “A little clean,” and “Not clean at all.” From the exit poll, we measure the voter’s confidence that her vote will be counted using the following question: “In general, how confident are you that the vote you cast for president will be respected and counted for the final result?” Respondents chose among the following options: “Very confident,” “Some- what confident,” “A little confident,” and “Not at all confident.”
The summary of findings are below. One note (not in the quote); the presence of election observers had no impact on perceptions of integrity.
On one hand, we show that confidence in the electoral process among supporters of the incumbent party decreased only after realizing that their candidate had lost. This change in the perceptions of electoral integrity responds to a pure “losers’ effect,” in which supporters of a losing candidate try to explain her defeat as a consequence of a poor electoral administration. On the other hand, we show that the discredit of electoral integrity among supporters of a party that has never won the presidential election is consistent over time. In this case, the skepticism from leftist partisans arose from both the systematic manipulation against left-wing parties during the twentieth century, and the dis- course of electoral distrust expressed by left-wing parties during recent presidential campaigns.
The full paper is available on early release.
Ok, this is just unfair:
At least he humored me by laughing at Doug Chapin’s “Nerd-Vana” joke.