A new article in the American Political Science Review by four graduate students at Harvard University uses a creative field experiment to show that local election officials are less likely to respond to informational inquiries from individuals with “putatively Latino names.”
In the article, titled “What Do I Need to Vote? Bureaucratic Discretion and Discrimination by Local Election Officials”, the authors describe the results of a large (N=6825) contact efforts, spread across 46 states. The emails contained requests for information about voting or about requirements for a voter ID and are fairly generic:
The text of the voter ID email was as follows:
I’ve been hearing a lot about voter ID laws on the news.
What do I need to do to vote?
(Jose Martinez, Jake Mueller, Luis Rodriguez, or Greg
The control email was as follows:
I’ve been wondering about this. Do you have to vote in
the primary election to be allowed to vote in the general
(Jose Martinez, Jake Mueller, Luis Rodriguez, or Greg
These are fairly
generic emails, but there was a statistically significant lower probability of receiving any response and receiving an informative response for those emails sent from names that appeared to be Latino. See the table for the key results (click on the image for a larger view).
The authors are quick to note that this is not an article about election officials per se, but about discretion provided to “street level bureaucrats” in implementing laws and regulations. However, they also note that this may raise concerns about the impact of voter ID laws on specific populations.
For interested readers, the full abstract is below:
Do street-level bureaucrats discriminate in the services they provide to constituents? We use a field experiment to measure differential information provision about voting by local election administrators in the United States. We contact over 7,000 election officials in 48 states who are responsible for providing information to voters and implementing voter ID laws. We find that officials provide different information to potential voters of different putative ethnicities. Emails sent from Latino aliases are significantly less likely to receive any response from local election officials than non-Latino white aliases and receive responses of lower quality. This raises concerns about the effect of voter ID laws on access to the franchise and about bias in the provision of services by local bureaucrats more generally.
A story in the San Jose Mercury News has the seemingly odd title: “Snail Mail the solution to slow Silicon Valley vote totals?”
And it is an odd story–the reporter is trying to get at a point but has botched it pretty badly under some inaccurate quotes and incomplete understanding about the technology of processing mail ballots and peculiarities of California election law.
Let’s start with the quote, about voting by mail and turnout. I’m not quite sure if Tony Green, chief spokesperson for Secretary of State Kate Brown, said what the reporter said he said–the one sentence lead in just doesn’t jibe with Tony’s quote:
Advocates say all-mail elections boost turnout while avoiding the equipment and personnel costs of traditional polling places.
“Oregon had the highest turnout of any state in the country in November,” said Tony Green, spokesman for that state’s elections office. He said they usually have very strong turnouts that are credited to the ease of voting by mail. “You have mailing costs, but significantly lower personnel costs.”
The quote is all about mailing costs and personnel–but there is that (accurate) statement that Oregon had the highest turnout of any state in 2014. Is this attributable to voting by mail? If it is, that would be something of a surprise, since I have shown in numerous places (most recently here with my colleague Peter Miller) that voting by mail makes a slight contribution to Oregon’s turnout rate–at best a few percentage points.
Oregon’s demographics are what mainly contribute to it being a high turnout state. I haven’t look in detail at the 2014 data, but I’d be very surprised if the comparatively high turnout in 2014 was because Oregon comparatively has lower percentages of those groups (notably African Americans and Latinos) who turned out at comparatively lower rates in 2014.
In addition, Oregon had a number of high profile ballot initiatives, into which outside donors sank tens of millions of dollars that smashed previous records for such spending.
I’ve had great interactions with Tony, as I’ve had with all of the election officials in Oregon, but the more accurate statement would be that VBM has led to clean registration rolls, lower costs, and high levels of voter satisfaction, even if it does not translate into substantial increases in turnout, at least in federal contests.
But the real gist of the story in the Merc has to do with how ballots are cast and counted in California. California has a very liberal set of balloting procedures, put in place so that no citizen is disenfranchised just because they happen to drop a ballot at the wrong precinct, or drop an absentee ballot at the precinct place, and now even if they mail the ballot by election day (rather than delivering it by election day).
(Kim Alexander of the California Voting Foundation details the various procedures in this extensive report.)
The main reason that California has a slow count is that a) millions of California voters hand deliver their absentee ballots to the precinct place on election day. These ballots are not read through the optical scanner at the precinct place–they can’t because the signatures need to be verified, the ballots need to be separated from the ballot envelope, the ballots need to be inspected (and potentially “remade” or “remarked”–and this can only be done under the scrutiny of an elections board), and, finally, the ballots can be counted.
The technology referred to by the reporter, in place in King County, WA, can not speed up the tallying of ballots in the Valley if and until California were to go fully by mail. The reason Washington, and Oregon, and Colorado, and other fully vote by mail states can process their mail ballots quickly is because all counting is done at the central office.
There are not tens of thousands of precinct places with millions of unprocessed vote by mail ballots accumulating in “red bags,” waiting to be delivered to the county offices after polls close.
All in all, the story is a mish mash. “Snail mail” really has little to do with the question of the slow count, and vote by mail won’t solve California’s slow count unless they get rid of precinct place voting altogether. I don’t see that on the horizon.
There has been a lot of ink spilled over a recent article in the Monkey Cage that suggested that “Non Citizens Could Decide the The November Election.” At last count, the post had generated 3305 comments, the most by far in the history of the Monkey Cage.
The blog posting was based on a forthcoming article in Electoral Studies, which had a less provocative title (“Do Non Citizens Vote in US Elections“) but does contain this highly charged claim:
These results allow us to estimate the impact of non-citizen voting on election outcomes. We find that there is reason to believe non-citizen voting changed one state’s Electoral College votes in 2008, delivering North Carolina to Obama, and that non-citizen votes have also led to Democratic victories in congressional races including a critical 2008 Senate race that delivered for Democrats a 60-vote filibuster-proof majority in the Senate.
For those who wish to further restrict participation by non-citizens, however, our results also provide important cautions. Simple resort to voter photo-identification rules is unlikely to be particularly effective.
But I have to add that this quote, in a recent “Fact Checker” article in the Reno Gazette-Journal, is just brutal.
UPDATE: After this story posted, Richman replied via email:
“We agree with your rating of a ’4′ because:
“A. Noncitizen voting might tip one or two extremely close races but is unlikely to tip the balance in the Senate, and certainly not in the House.
“B. Science is a process of finding, validation, replication and rebuttal. We are at the very beginning of the process. Colleagues have raised reasonable questions about the data we used–problems that we acknowledge in both the study and the Monkey Cage. It will take some time and additional research to increase confidence in our findings.”
Horse. Stable Door. Too Late.
The damage form this study may have already been done. Doug Chapin, someone who bridges political science and policy, has already written (“Is Political Science Blowing It’s Close Up?”) about the impact of this study (and the Montana experiment) on when and how election administrators may engage with scholars. I am attending a conference of election officials in just a few weeks, and I am certain I will have to defend our discipline from those who are already skeptical about working with scholars.
Any political scientist, and particularly those who work in the elections administration and election policy fields, need to be worried to see a quote like this from one of our supporters and friends:
But if political scientists aren’t careful – either in monitoring their own or their colleagues’ research and publishing decisions – the interest in political science-driven stories will wane. Or worse, it could become yet another (albeit more numerate) weapon in the ongoing rhetorical wars between the parties.
It will also make it harder for researchers and election officials to “play nice” with one another on projects of mutual interest – which for me would be the unkindest cut of all.
My professional association is working hard to convince politicians and policy makers that our scholarship can be relevant. But we as members need to be very circumspect about how we publicize our work, particularly in the context of a dynamic and competitive election campaign. This is not about a few citations or a few appearances on local news shows. This is about political power, and those in power can be quite unforgiving.
More on GlacierGate: If non-partisan judicial elections are not a democratic good, can we conduct research intended to actively undermine them?
Thomas Leeper, in a recent blog posting, makes what strikes me as a very problematic claim to try to justify the Montana field experiment.
Leeper asserts that non-partisan elections “do not obtain the democratic benefits that their advocates hope for,” and that “judicial elections are not necessarily a democratic good.”
I defer to Prof. Leeper for the justifications of these claims; I have no reason to doubt his summary of the literature. I find his arguments intuitively and theoretically appealing.
But how can this possibly justify the Montana field experiment? Leeper is arguing that scientific research that in the process of conducting the research actively undermines a democratic election practice cannot be criticized if the process itself is of questionable democratic value.
Please note, I am not saying that political scientists should not subject election procedures to the closest possible empirical and normative scrutiny. But Leeper misses the point, made by myself in an earlier post and by Melissa Michelson on the New West Blog, that this experiment did not just study the impact of providing partisan cueing information on voter turnout in a non-partisan election, by its very scope, could have undermined the practice itself.
There are 671,031 registered voters in Montana, so this mailer was sent to 15% of the electorate. Depending on how many of the recipients had already intended to vote, using the 2010 turnout as a baseline, as much as half the total voting population received this mailer!
Choose your guide to research ethics in the social sciences. Here is one from Notre Dame, and second from Iowa State. I didn’t choose these with any particular intent in mind; they were just two of the first that came up after a Google search of “ethical guidelines for social science research.”
Others may disagree, but I fail to see how this study attempted to, at a minimum:
- Consider and anticipate effects on third parties that are not directly included in the research (judicial candidates, supporters of non-partisan elections in Montana)
- Show respect for the values and views of research subjects, even if they differ from those generally accepted by society at large (if we accept Leeper’s argument that non-partisan elections are a net bad, and so if the experiment undermined the Montana election it’s OK since those who believe this are simply wrong)
The example used in research ethics 101 is this: we cannot be absolutely sure that someone does not have HIV (today the example used would be Ebola) unless we tested all of their blood. The problem with this test: it would kill the individual. We should minimize to the degree possible the impact of our measurement on the thing we are measuring, and this research design fails this test.
Finally, I’m really amazed that this research is justified on the grounds that private entities are doing this anyway. John Patty writes:
I will point out quickly that this type of experimental work is done all the time by corporations. This is often called “market research” or “market testing.” People don’t like to think they are being treated like guinea pigs, but trust me…you are. And you always will be.
Corporations are not subject to an IRB. I we hold ourselves to a higher standard than simply what makes money for Anheuser Busch.
A provocatively titled posting at the Monkey Cage suggests that Non Citizens Voting Could Decide the 2014 Election.
I discussed the Electoral Studies article that the Monkey Cage posting is based on at Early Voting.net, and expressed concerns then that the article made a number of very heroic assumptions to be able to claim that non-citizens were voting in significant numbers, and even more heroic assumptions to assume that these votes “created the filibuster proof majority in 2008,” as the authors claim.
Now the authors have doubled down, writing on Monkey Cage that non-citizens “could decide” the 2014 election, whatever that means in the context of House, Senate, gubernatorial, state legislative, and other races.
I’m engaged with my professional association in trying to show the public relevance of political science, but this isn’t exactly what I had in mind. There are heated public debates going on right now about the voter identification, and regardless of which side of this debate you are on, it’s dangerous to inject yourself into this debate based on a first look at a question like this, based on what many other scholars consider to be extremely tenuous assumptions.
Rick Hasen has posted a very nice followup on his blog that summarizes the situation far better than I can.
For those readers interested in the more detailed criticisms,
I’ve provided a link to the whole thread from the Election Law listserv. (Against listserv policy, apologies to fellow list members. Suffice it to say that there are trenchant criticisms, and I’ve encouraged those posting to enter the public dialogue.)
I encourage readers to pay especially close attention to any critiques provided by Michael McDonald. McDonald is the expert on identifying the number of non-citizens among the population, an exercise he engages in every two years in order to produce his estimates of the voting age population (VAP), voting eligible population (VEP), and voter turnout.
This one is not over, I am sure of that, and I expect to see additional scrutiny and replications in the next few months. This will not be soon enough to avoid inevitable post-election charges that in-person voter impersonation is rampant.