“My father will accept it 100 percent if it’s fair — if it’s fair. . . . Fourteen percent of all noncitizens in this country are registered to vote.”

— Eric Trump, interview on ABC’s “This Week,” Oct. 23

After refusing to commit to accepting the results of the presidential election, Republican nominee Donald Trump now says he would accept them — under certain conditions. He said that he would accept a “clear election result,” and that he would challenge a “questionable” one. And he said he would accept the results if he wins.

Eric Trump, one of the nominee’s three sons, repeated his father’s stance in an interview. He cited a series of Four-Pinocchio claims to show what an “unfair” election could mean. He then said that “14 percent of all noncitizens” are registered to vote, which is a version of a claim we debunked last week and is worth explaining again.

The Facts

This figure comes from research that was published two years ago in the Monkey Cage, a blog hosted by The Washington Post. The Trump campaign is using this research to falsely claim that illegal immigrants are voting in and tipping elections.

Old Dominion University professors Jesse Richman and David Earnest studied voting participation rates of noncitizens by using data from 2008 and 2010 collected through the Cooperative Congressional Election Studies. This data set comes from a YouGov-Polimetrix opt-in Internet survey. Researchers were able to cross-check 40 percent of the data they collected from 2008.

Based on results from 339 noncitizen respondents in 2008 and 489 in 2010, researchers found that more than 14 percent of noncitizens in the 2008 and 2010 samples said they were registered to vote.

Eric Trump goes even further, applying this statistic to the general noncitizen population. If his claim were accurate, then about 3.2 million of the estimated 23 million noncitizen population would be registered to vote. But researchers warned that “it is impossible to tell for certain whether the noncitizens who responded to the survey were representative of the broader population of noncitizens.”

A number of researchers were skeptical of the findings and methodology, and they published critiques of it. Some critiques are now being incorporated into a revision of the original study.

The original column on Monkey Cage includes this editor’s note at the beginning of the article: The post occasioned three rebuttals (here, here and here) as well as a response from the authors. Subsequently, another peer-reviewed article argued that the findings reported in this post (and affiliated article) were biased and that the authors’ data do not provide evidence of non-citizen voting in U.S. elections.

Stephen Ansolabehere, who created and runs the Cooperative Congressional Election Study (used in the research by Richman and Earnest), and two other researchers offered the most thorough critique. They replicated the research by interviewing the same panel of respondents about their voting patterns in 2010 and 2012. They found that people had identified as a citizen one year, but noncitizen the next — indicating misclassification.

They concluded that all of the cases of noncitizen voting “are nearly certainly citizen voters who are misclassified as being noncitizens. Hence, their predicted vote rates of noncitizens in fact reflect the behavior of citizens.” Because the citizen group is large compared with the noncitizen group in the survey, even a small classification error could be substantial, they found.

Ansolabehere responded to Trump’s comments about noncitizen voters in an Oct. 19 Monkey Cage post, recapping his critique “that this study is wrong and that there is absolutely no evidence from the data that non-citizens voted in recent presidential elections.”

Richman said that Ansolabehere’s critique was thoughtful and useful, and that he is working on a revision based on the critiques raised in the response. But Richman said his results are valid and rejected claims that his findings were entirely spurious. Richman wrote a response to the use of his data by the Trump campaign, clarifying that his research does not support claims of massive voter fraud by illegal immigrants.

The Pinocchio Test

If a statistic sounds too fantastic to be true, then it’s probably false.

Eric Trump uses a statistic from widely criticized research to make a point that “unfair” election practices are taking place. He asserts that 14 percent of noncitizens are registered to vote, but this is an unreliable data point from a disputed sample and is not representative of the larger noncitizen population. It’s pretty shocking that he would seize on a disputed statistic from a small sample — and then apply it to the entire country. We award Four Pinocchios.

Four Pinocchios

Send us facts to check by filling out this form

Sign up for The Fact Checker weekly newsletter