A controversial 2014 study used survey data to demonstrate that 38 world might have voted as non-citizens in 2008, and web sites ultimately did a many extrapolating.

You are watching: How many illegal immigrants voted in 2012

Published23 June 2017

Share on share on facebook on TwitterShare on PinterestShare ~ above RedditShare via Email


Credible evidence says that 5.7 million illegal immigrants can have poll in the 2008 election.

On 20 June 2017, the Twitter account of the morning show “Fox & Friends” tweeted a significant sounding little bit of news:

As many as 5.7 million illegal immigrants could have poll in the 2008 election, report finds.

A much more accurate tweet, if it might fit, could be:

The Washington times is reporting that a net site named “JustFacts.com” has actually concluded that a widely-discredited 2014 study suggesting up to 2.8 million non-citizens vote in the 2008 presidential choice (based ~ above the extrapolation of 38 survey responses from people who may have actually voted together non-citizens) has been unfairly debunked by “liberal reality checkers” and that, in reality, the number might be together high as 5.7 million.

The original Study

That 2014 study, released in the journal Electoral Studies and authored through Jesse Richman, Gulshan Chattha, and David Earnest at Old preeminence University, made waves once the researchers very first described their results in a Washington post column that influenced three different rebuttals — and one added rebuttal to those rebuttals, as well as a disclaimer around the discussed nature of the research file itself.

In this study, the authors provided data gathered by internet polling firms because that a Harvard university initiative known as the participating Congressional choice Studies, or CCES:

The 2008 and 2010 participating Congressional Election research studies (CCES) were carried out by YouGov/Polimetrix the Palo Alto, CA together an internet-based survey using a sample selected to mirror the demographic qualities of the U.S. Population.

In both years survey data was collected in 2 waves: pre-election in October, and then post-election in November. The questionnaire asked more than 100 questions about electoral participation, worry preferences, and candidate choices.

The thrust of their work-related was to demonstrate that some human being checked off the they to be both non-citizens and that lock voted, in some cases going so far as to describe the candidate castle voted for. As a inspect of their work, they offered information provided by CCES indigenous a study firm called Catalyst come verify that world who said they voted in reality voted:

Validation that registration and also voting to be performed by the CCES research study team in collaboration with the certain Catalyst. Of 339 non-citizens determined in the 2008 survey, Catalyst matched 140 come a advertising (e.g. Credit card) and/or voter database.

Out the the 38 cases from 2008 in i m sorry non-citizens claimed to have actually voted (or had actually a vote validated they didn’t recognize to in the survey), the authors found five (as in, the number after ~ four) situations of survey responses native non-citizens who both said they had voted and that Catalyst can verify as having actually voted.

Using this data, part modeling, and error analysis, the authors concluded that in between 7.9 percent and 14.7 percent of non-citizens poll in the 2008 elections. They climate simply applied this come the entire non-citizen population in the joined States. The findings space as crude as they space controversial:

Since the adult noncitizen population of the United states was around 19.4 million, the number of non-citizen voters <…> could range from just over 38,000 in ~ the an extremely minimum to nearly 2.8 million in ~ the maximum.

These numbers remainder on the presumption that a subset that 38 (possible) non-citizen votes out of 339 non-citizens can be provided to extrapolate countrywide poll behavior.

The Rebuttal

If extrapolating to a number based from net survey an answer data native a pool of 339 non-citizens right into the millions sound problematic come you, you space not alone. Brian Schaffner is professor that political scientific research at the college of Massachusetts, Amherst and the co-principal investigator that the Harvard CCES from which Richman obtained his data. He told us via e-mail:

I nothing know any kind of serious survey researchers who would have tried to extrapolate 100 or so respondent from a big survey favor this to produce a range that huge without tracking ago to think around the dubiousness of the projection. <…>

It is totally worthless as a selection of anything.

Schaffner was an author on a difficulty to the Richman paper (“The dangers of Cherry Picking short Frequency occasions in large Sample Surveys”), likewise published in Electoral research studies in 2015. Schaffner’s paper makes the dispute that also a nearly non-existent amount of misreporting native the non-citizen group would create deeply flawed outcomes if one do the efforts to usage that data come extrapolate. In the paper, they market the following mental exercise:

Suppose a survey inquiry is inquiry of 20,000 respondents, and that, of this persons, 19,500 have a given characteristic (e.g., are citizens) and also 500 perform not.

Suppose that 99.9 percent of the moment the survey inquiry identifies properly whether people have a given characteristic, and 0.1 percent of the time respondents who have a offered characteristic mistakenly state the they carry out not have that characteristic. (That is, they inspect the wrong crate by mistake.)

That means, 99.9 percent of the moment the question properly classifies an individual as having a characteristic — such as being a citizen of the United states — and 0.1 percent of the time it classifies someone together not having a characteristic, when in fact they do. <…> the implies, however, the one expects 19 civilization out the 20,000 to be erroneously classified as not having a provided characteristic, once in truth they do.

Suppose that 70 percent the those with a provided characteristic (e.g., citizens) interact in a habits (e.g., voting). Suppose, further, the none of the human being without the characteristic (e.g., non-citizens) are enabled to interact in the behavior in question (e.g., poll in federal elections). Based upon these suppositions, of the 19 misclassified people, we mean 13 (70%) to it is in incorrectly identified to it is in non-citizen voter while 0 effectively classified non-citizens would be voters.

Hence, a 0.1 percent price of misclassification <…> would certainly lead researchers to intend to observe the 13 the 519 (2.8 percent) human being classified together non citizens vote in the election, once those results are due completely to measure up error, and no non-citizens in reality voted.

To more raise the possibility that this kind of error might have happened and could be significant, Schaffner and his partner went earlier and re-interviewed civilization in the survey using data native 2010, informing us:

In 2012, we re-interviewed 19,000 world who had actually been respondents for the 2010 CCES. We asked them the very same question around citizenship condition as we had asked castle in 2010. Of this 19,000, 121 had asserted to be non-citizens in in 2010. In 2012, 36 that the 121 had adjusted their an answer and to “citizen.”

Additionally, 20 civilization who had clicked on the “citizen” alternative in 2010 adjusted to “non-citizen” in 2012. Thus, that is plainly the instance that a tiny share of respondents were mis-clicking on response options to that inquiry in at least one the the 2 surveys (about .3 %).

The existence of also the possibility of misreporting, especially when you take into consideration that only 5 (5) that the non-citizen voters identified in 2008 were actually proved as voting, is problematic, as articulated by university of California, Irvine political scientist Michael Tesler in his Washington write-up rebuttal to the Richman study:

With the authors’ extrapolations of the non-citizen voting population based on a small number of validated votes from self-reported non-citizens (N = 5), this high frequency of an answer error in non-citizenship standing raises important doubts about their conclusions.

The Washington Times / “Just Facts” Take

One surefire method to make it sound like something carries government without in reality understanding any aspect the the topic you are covering would certainly be to describe the process, together the Washington times did in the story attached by “Fox & Friends”, as “a series of facility calculations”.

Outside that the truth that these calculations are discovered in the 1,010th footnote that the JustFacts.com report, the calculations (shown below) don’t involve lot more complex mathematics than multiplication, subtraction, and enhancement (no division, thankfully). What simply Facts did was take the United claims Census Bureau estimate of the variety of non-citizen adults in the United says (19,805,000) and also multiply that by, in essence, high-end and low-end estimates of the percent of people in that group who vote in elections based on data from the Richman study — but with their own estimates of error:

19,805,000 non-citizen adult × ((8% self-declared vote – 5% margin of error) + (8% undeclared vote – 8% margin the error)) = 594,150

19,805,000 non-citizen adults × ((8% self-declared voting + 5% margin that error) + (8% undeclared vote + 8% margin that error)) = 5,743,450

The “8% self-declared voting” number comes from the 27 non-citizens the end of 339 in the Richman study who said “I definitely voted”. The “8% undeclared voting” also comes from the same study, and is calculated together the 11 non-citizens determined by the Catalyst system as voting (out that the total 140 verified non-citizens matched to records in the Catalyst database). Any conclusion around sweeping waves of countless non-citizen votes is bound to this undeniably tiny numbers.

In a 15 December 2016 post, JustFacts.com’s president James Agresti noted its justification for taking the results of the Richman research seriously. This post, however, serves largely as an initiative to debunk the insurance claim made by Schaffner and his colleagues in their 2015 paper that “zero” non-citizen votes were actors in the 2008 presidential election.

For his part, Schaffner told us:

What we are saying <…> is that as soon as you account because that measurement, the best estimate that the number of non-citizen voter is zero.

That doesn’t mean we in reality think there room zero non-citizen voters.

The JustFacts.com post additionally does very small to attend to the truth that the Richman study’s non-citizen dataset was so limited:

The movie critics make a legitimate suggest that random errors by survey respondents will certainly overcount non-citizens. This is because far more citizens were sampled in this survey. Because that instance, if a inspection sampled 100,000 citizens and 100 non-citizens, and also 1% of castle misidentified themselves, this would average 1,000 citizens called themselves non-citizens, but only one non-citizen claimed he to be a citizen.

Such logic provides sense in a vacuum whereby all other proof is ignored, yet the fact is that misidentification that citizenship is not just a arbitrarily phenomenon. This is because illegal immigrant often claim they are citizens in order to conceal the fact that they space in the U.S. Illegally.

Agresti support the latter part of this explain by providing evidence that “certain groups of illegal immigrants” frequently use fraudulent Social security numbers and also “misrepresent themselves together citizens”. That brushes off the former part of the explain by echoing claims made by Richman in a working paper (not peer-reviewed) that various other demographic data in the CCES, and their own investigation of voter it is registered data, prove that civilization were no misreporting your citizenship standing after all.

People can dispute the virtues of those disagreements as much as lock want, but they certainly do not prove that zero civilization misreported their citizenship status, or that millions of non-citizen votes emerged in the 2008 election. The arguments additionally do not readjust the fact that the conclusion that “5.7 million noncitizen votes in 2008” is based on applying vast estimates of behavior from an exceedingly tiny subpopulation. The difficulties with this approach are confirmed by the absurdly huge possible range Richman and also later Agresti jointly came up v from the very same data (38,000 to 5.7 million illegal votes).

We request Richman exactly how he felt about Agresti’s analysis of his work, and also his response concedes the point that there is a the majority of room come play roughly with this kind of data:

Ultimately there are a selection of assumptions one can make as soon as interpreting the survey data, so ns am no surprised the a various analyst approaching the numbers with a different collection of assumptions might pertained to a patent different collection of conclusions. My impression is the this is what Mr. Agresti has actually done. And also while those numbers room not the persons I came to, at some point it comes down to which collection of assumptions one think are many plausible.

Straight-faced insurance claims that there is material proof for up to 5.7 million non-citizen votes in 2008 room remarkable, offered that the study frequently cited together the basis because that this case has listed material proof for five (not even six!) non-citizen votes in that year.

Any analysis based turn off of Richman’s study or Agresti’s analysis must square itself with the reality that lock are based upon numbers generated from just these facts:

1) In a group of 339 self reported non-citizens, 27 insurance claim to have actually voted; and

2) In a group of 140 confirmed non-citizens, 11 may have voted.

See more: How Much Did Jake Paul Make Boxing, Jake Paul Vs

In the defense of “Fox & Friends”, we identify that this added information would certainly make for a much less flashy tweet.