The evidence available, drawing from extremely little legal enforcement action (mostly tied to local scandal) and from a recent study (by voting fraud allegers) of Virginia, is that the number of non-citizens voting is infinitesimally small, under one tenth of one percent of actual voters, and probably that is a gross over-estimate. In contrast, the allegers of voting fraud are saying that somewhere around 10% of non-citizens voted and that somewhere around 2% of all voters were non-citizens.
For my state of Vermont, which is subject as all other states to Kobach’s demand for voter data, a high estimate of non-citizens registered (per the analysis below) is a total of 39.
Background: Kobach
Kansas Secretary of State Kris Kobach, who co-chairs the Trump Administration’s Presidential Advisory Commission on Election Integrity (VP Pence is chair) announced on November 30, 2016 that it was a “reasonable estimate” that 3.2 million people could have voted illegally based off a survey of the 2008 presidential election.
The Kansas secretary of state said data showed that 11.3% of non-citizens in the United States said they had voted in that year’s election. He pulled this figure from the heavens. There are 43 million immigrants of which 95% are 18 and older . About half are naturalized. Kobach is in effect alleging that of about 20 million non-citizens 18 or old, 11.3% or 2.3 million voted illegally. (If all of Kobach’s 3.2 million voted illegally and were non-citizens 18 or over, that implies that 16% of non-citizens voted.
On the face of it, this allegation implies national. massive, orchestrated campaigns of voter fraud. Actual systemic voting fraud cases of late involve at most dozens of voters and are local in nature.
A disputed 2014 research article
Kobach and allies have leaned on a 2014 article authored by Old Dominion University researchers that “6.4 percent of non-citizens voted in 2008 and 2.2 percent of non-citizens voted in 2010.”
The article was based on no original inspection of records or surveying but rather on a study by the Cooperative Congressional Election Study of surveys done in 2008 and 2012. The CCES researchers disparaged the Old Dominion article asserting that is drew unwarranted inferences from a very small sample size (such as under 200 positive results from a total survey population of 19,000). The title of the CCES December 2015 refutation: “The perils of cherry picking low frequency events in large sample surveys.”
The Virginia allegations
In May of this year a Public Interest Legal Foundation – sponsored report, “Alien Invasion,” produced figures of what it called non-citizens with Virginia driver licenses who voted. If you examine its figures, and compare them with Virginia population and voting numbers, these non-citizen estimates come to one third of one percent of adult non-citizens, and an infinitesimally small percentage of total voting in Virginia.
The Virginia report compared motor vehicle registration data (which has a field for citizen status), registration rolls, and actual voter counts. I have grossed up the figures, which covered most but not all voter districts, in order to show complete statewide estimates. The adjusted figures indicate that over a six year recent period 2,415 non-citizens were registered to vote and that they voted 9,745 times. These are definitely not the figures that Kobach and allies want to hear about. The reason is that the 2,415 is an extremely tiny share of the approximately 400,000 non citizen adults living in Virginia at the time.
The reasonable explanation is that many or most of the 2,415 persons either became citizens later or mistakenly listed themselves as citizens (consistent with survey errors found by CCES). Also, the 9,745 votes over a six year period is a vanishingly small share of the roughly 18 million times people voted in Virginia over the six year period.