Several internet headlines, on the same topic, caught my eye yesterday. One, from townhall dot com, read “Oh My: Study Reveals Significant Number of Non-Citizens Vote in US Elections,” by Guy Benson, a conservative writer with a bachelor’s degree from Northwestern..
The “study” to which the articles refer was published in 36 Electoral Studies 149-57 (2014) and was written by professors (Jesse T. Richman, Gulshan A, Chattha and David C. Earnest) from Old Dominion University and one from George Mason University, and is titled: “Do non-citizens vote in U.S. elections?”
At first, the “study” seems mind-blowing. In Benson’s column, in bold print (provided by him or someone else; he does not indicate), the study in question has big numbers: “Its large number of observations (32,800 in 2008 and 55,400 in 2010) provide sufficient samples of the non-immigrant sub-population, with 339 non-citizen respondents in 2008 and 489 in 2010.”
Anyone who has taken a course in statistics knows the value and importance of sample size. Benson quotes some of the inferences derived by the “study”: that in 2008 Al Franken might have won his Senate seat in Minnesota on the strength of votes by non-citizens, and thus Obamacare would not have passed a filibuster-proof Senate.
I ordered a copy of the Electoral Studies article. Nine pages cost $19.95. The conclusions drawn by the professors bear little legitimacy and the hype of the headlines should be disregarded. Here are several things one should know about the “study.”
1) The article itself is not a “study.” A study is a survey of people with controls in place and checks on interpretation of data derived. Rather, the article is the authors’ interpretation of numbers from two (2) years’ results of Cooperative Congressional Election Studies. (Richman, p. 150.)
2) The numbers of the sample size are big, but only as to overall samples. “The 2008 and 2010 [CCES] were conducted by YouGov/Polimetrix of Palo Alto, CA as an internet-based survey using a sample selected to mirror the demographic characteristics of the U.S. population.” (Id.) The article is based on a survey. The purpose of CCES, according to a website titled under its name: “Broadly, the Common Content consists of a battery of questions asked of all respondents to capture commonly asked questions such as vote choice, as well as a handful of items for which it is uniquely advantageous to have a very large sample.”
3) Unfortunately, the sample upon which the authors depend, is quite small. While one might be impressed by 32,800 respondents in 2008 and 55,400 respondents in 2010, the conclusions Richman, et al., derive are from the “non-citizens” who responded to the CCES: 339 in 2008 and 489 in 2010. In study of statistics in political surveys, we were taught a sample number under 900 was statistically suspect.
4) The numbers shrink from there. Of “339 non-citizens identified in the 2008 survey, [the firm Catalyst] matched 140 to a commercial (e.g., credit card) and/or voter database.” (Id.)
5) These questionable aspects of the results of Richman, et al., are reflected in some of the statements in their article:
-Perhaps most critically: “A critical question for this project is whether respondents’ self-identification as non-citizens was accurate. If most or all of the ‘non-citizens’ who indicated that they voted were in fact citizens who accidentally mistated their citizenship status, then the data would have nothing to contribute concerning the frequency of non-citizen voting.” (Id.) If I received a CCES questionnaire, I would find offensive someone asking me how I voted, my views on specific issues, and whether I am a United States citizen. I have self-identified, in such surveys years ago before I was a lawyer, as belonging, amongst other categories, to “non-English-speaking Aleuts.” As a member of one household in which I lived, I identified a Persian cat as an Iranian immigrant without papers. I do not want to provide institutions with information upon which those institutions can attempt to manipulate the vote. I do that enough with purchases, especially the Bill Maher Fan Club coffee mug and t-shirt I bought.
-These problems do not deter Richman, et al, who confidently footnote their article: “Since the total legal permanent resident population in 208 of 12.6 million ... was approximately four percent of the overall U.S. population, and the total non-citizen adult population in 2011 was 19.4 million ... the non-citizen population was under-sampled. Nonetheless, the sample that was collected provides the first nationwide sample from which analysts can draw inferences concerning electoral participation by non-citizens in United States elections.” Richman, p. 150, n. 2. What? That means nothing. There were a lot of people in the United States, a lot of people were asked questions in the survey administered by other people, and the authors could extrapolate from those results.
-The real cherry-picked numbers of people from whom results are extrapolated are, for example, 67 in 2008 who “either claimed they were registered [to vote] had their registration status verified, or both.” (Richman, p. 151.) If more than a few statistics merrymakers responded to the survey, that “67" is quite suspicious. Even then, the authors qualify it by saying some were verified.
-The gem of the article is: “The ‘adjusted estimate’ row presents our best guess as the true percentage of non-citizens registered [to vote.” (Id.) That is a very scientific concept. The authors provided us with their “best guess.” Whoops—they use the same phrase a second time. (Richman, p. 152.) I guess that means there are “gems” in the article.
-The authors go all-out on inferences: from 56 non-citizens who indicated they did not vote, 13 who indicated they voted and five whose vote was validated, the “adjusted estimate of 6.4 percent for 2008 is quite substantial, and would be associated with 1.2 million non-citizen votes cast in 2008 if the weighted CCES sample is fully representative f the non-citizen population.” (Richman, p. 153.)
6) We can take some comfort from another footnote: “One important caveat is in order. To the extent that non-citizen voting is dependent upon an ability to ‘pass for’ a citizen at the polling place respondents who looked less like immigrants to election officials might have an easier time voting.” (Richman, p. 155, n. 8.)
This article advances conclusions with little validity. We can be certain the article will receive even more publicity than it has in the past two days. We also can be pretty sure the authors received grant money to write the piece, and that they will list its authorship on their resumes when they apply for other grants, or perhaps for faculty positions at Pat Robertson’s Regents University ot the late Jerry Falwell’s Liberty University, or that bastion of free thought and advancement of racial equality, Bob Jones University.