In criticizing Donald Trump’s unsupported claim that millions of noncitizens swung the popular vote, the media have gone too far in the other direction, dismissing legitimate — albeit merely suggestive — evidence that significant noncitizen voting does occur in American elections. Take the much-discussed study [pdf] by Jesse Richman, Gulshan Chattha, and David Earnest, published in the journal Electoral Studies. Based on a national survey, the authors found that a small percentage of noncitizens do report voting — roughly 6 percent in 2008 — and that their votes might be enough to affect very close elections, such as the Minnesota Senate race between Al Franken and Norm Coleman.
The Richman-Chattha-Earnest study was immediately criticized for overinterpreting what might be just measurement error in the citizenship variable. (In other words, citizen voters may have accidentally labeled themselves noncitizens in the survey.) But subsequent robustness tests by the authors show that at least some of the self-identified noncitizen voters in the authors’ dataset really do appear to be noncitizens. I looked into this debate in some depth recently, and I came away feeling that there is still too much uncertainty to draw strong conclusions, but the Richman-Chattha-Earnest study is certainly a valuable contribution.
At Politico, Brent Griffiths apparently disagrees. After a Trump spokesman cited the Richman-Chattha-Earnest study, Griffiths reported it as “debunked” and “refuted,” without mentioning the authors’ responses to the supposed debunking. He even incorrectly characterized their study as focusing on illegal immigrants, when the subjects of the study (and of Trump’s claim) were noncitizens in general.
To its credit, the Washington Post fact-checker at least quotes Jesse Richman, the first author of the study, and gives his side of the story. Nevertheless, the Post warns that the Trump team must stop citing the study — lest Trump draw the fact-checker’s cruel wrath again, presumably.
There are two lessons here. First, politicians and media figures must do a better job of accepting uncertainty. If there is any overarching theme to my writing, it is that people are much too confident in their beliefs about the world. Advocates will seize the mantle of Science with the flimsiest of evidence, then quickly dismiss contrary data as “debunked.” Sooner or later, a false consensus develops around ideas that were never firmly established — in this case, that illegal voting is nothing to worry about. In their rush to shout “Pants on fire!” the media seem to have forgotten the value of admitting “We just don’t know.”
The second lesson is that liberal dominance of the social sciences continues to pay dividends in the media. Once the Richman-Chattha-Earnest study was released, it met immediate hostility from experts intent on debunking it. This seems to happen whenever any academic study provides political ammunition (intentional or not) to conservatives. And yet the media landscape is suffused with junk science purporting to prove all sorts of liberal political claims: Diversity makes people smarter, same-sex marriage improves public health, only 2 to 8 percent of rape accusations are false, and so on. The media often treat these dubious claims as factual, but who can blame them when there is no army of debunkers they can learn from? I don’t pretend to know how to fix this imbalance, but keep it in mind the next time someone cites a “consensus.”
PS. Here’s a message for the next Census Bureau director. A simple tweak of the Current Population Survey could shed a lot of light on noncitizen voting. The November supplement asks respondents whether they are registered and whether they voted in the most recent election. Unfortunately, it asks those questions only of respondents who have already said they are citizens. Ask about voting before establishing citizenship, and let’s see what that turns up.