Illegal Voting By The Numbers

This post originally ran 31 January 2017. Voter fraud is an old, old story. Old posts do NOT re-email to followers.

How many votes in the past presidential election were cast illegally or fraudulently? Some say none to few. Others, such as President Trump, say a couple of million. The mainstream press insist there is “no evidence” for systemic problems in the electoral process. Yet evidence does exist, only that evidence is disputed or ignored.

Here is the story so far. President Trump lost the popular election by more than two million votes, apparently due largely to the massive number of blue votes concentrated in California and New York. Yet shortly after the election, he said, “In addition to winning the Electoral College in a landslide, I won the popular vote if you deduct the millions of people who voted illegally.” He later estimated the fraudulent margin to be some 3 to 5 million.

He also said, “I will be asking for a major investigation into VOTER FRAUD, including those registered to vote in two states, those who are illegal and even, those registered to vote who are dead (and many for a long time). Depending on results, we will strengthen up voting procedures!”

There has been no systematic attempt to estimate the illicit vote count, and such an estimate is required before dismissing or accepting Mr. Trump’s claims.

Mr Trump’s claims launched waves of horrified apoplexy in the press (who at this early point know no other reaction). The New York Times was reduced to using the L-word, i.e. “liar.” The Washington Post said Trump’s charge “is not supported by any verifiable facts.” Even politicians in his own party, such as Senator John McCain, said, “I obviously have seen no evidence of illegal voting.”

Various rhetorical tricks then played out in the mainstream press to give the impression illegal voting was rare in the extreme, or even non-existent. An academic study which estimated there were some 800,000 illegal votes from non-citizens was excoriated. Much evidence in plain site was just plain ignored. Yet, so far, there has been no systematic attempt to estimate the illicit vote count, and such an estimate is required before dismissing or accepting Mr. Trump’s claims. The best that can now be said is nobody knows the right answer.

Sloppy “Fact” Checking

One writer at the Washington Post “combed through the news-aggregation system Nexis to find demonstrated cases of absentee or in-person voter fraud.” And since this reporter could only find four reported instances, he said, “There is simply no evidence that fraudulent ballots played any significant role in the 2016 presidential election whatsoever.”

Arguing that since newspaper accounts of fraud don’t exist that therefore actual fraud doesn’t exist is not unlike arguing that since Pravda didn’t print reports of arrests of political prisoners in Moscow under Stalin, that therefore the arrests didn’t happen.

But since the point is question is fraud that has (thus far) gone undetected, arguing that since newspaper accounts of fraud don’t exist that therefore actual fraud doesn’t exist is not unlike arguing that since Pravda didn’t print reports of arrests of political prisoners in Moscow under Stalin, that therefore the arrests didn’t happen.

The controversial site FactCheck.org produced, at times, a petulant report, calling Trump’s claims “bogus.” FactCheck.org relied in part on the paper “The Truth About Voter Fraud” published by the left-leaning Brennan Center for Justice. It is a remarkable document that takes great pains to suggest that fraud almost never occurs by emphasizing instances where fraud was searched for but was not found. For instance, this bullet point: “In Washington in 2005, an individual asked county offices to investigate the citizenship status of 1,668 registered voters based on their ‘foreign-sounding names.’ There are no reports of which we are aware that any individual on the submitted list was actually a noncitizen.” This is clever because it also brings with it the slight whiff of racism, for which there is no defense.

The Brennan report argues repeatedly that those who break the law to come to the USA would not likely break the law a second time to vote because the “payoff,” i.e. their one additional vote, is so small. But that same logic (as is well known) applies to the legal citizen voter deciding whether to head to the polls knowing his one vote also counts for almost nothing in a general election.

Pew and Actual Fraud

FactCheck.org admits Mr. Trump quotes accurately from the Pew Report “Inaccurate, Costly and Inefficient: Evidence That America’s Voter Registration System Needs and Upgrade“, before downplaying the report because it doesn’t specifically mention fraud. Instead, Pew says things like “Approximately 24 million — one of every eight — voter registrations in the United States are no longer valid or are significantly inaccurate,” “More than 1.8 million deceased individuals are listed as voters,” and “Approximately 2.75 million people have registrations in more than one state.” While none of these are direct indicators of fraud, all are in the direction of fraud.

FactCheck.org does bring up an instance of fraud:

After the 1982 election, in Chicago, 62 people, most of them precinct captains, were indicted by a grand jury for stuffing ballot boxes and buying votes, including a scheme in which they would identify registered voters not voting on Election Day and forge ballots in their name.

Like certain other metropolitan areas, Chicago is, as all know, a Democratic machine town with a long history of electoral shenanigans. In the linked example, the Chicago Tribune reported that a Democratic precinct worker was caught taking a straight-Democrat ballot and running it through a vote-counting machine 198 times. In that same article, “U.S. Atty. Dan Webb repeated contentions previously made by federal investigators that of the 1 million votes cast in Chicago in the Nov. 2, 1982, general election, about 100,000, or 10 percent, were fraudulent.”

Also: Webb “estimated that 80,000 illegal aliens are registered to vote here” and that by that time some had already been convicted for fraudulent voting. Several election officials were also convicted and others awaited trial.

FactCheck.org then pooh-poohs these facts by reminding the reader that “Trump urged his supporters to ‘watch your polling booths…,’” thus suggesting cheating could not have occurred under these watchful eyes, and by citing “experts” who say “the kind of voter fraud Trump is talking about — voter impersonation — is extremely rare.” Rare is might be, but did it happen in 2016?

Major Malfunctions

FactCheck.org forgot (somehow) to mention headlines like this: “Voting machines in more than one-third of all Detroit precincts registered more votes than they should have during last month’s presidential election.” The main discovery: “Detailed reports from the office of Wayne County Clerk Cathy Garrett show optical scanners at 248 of the city’s 662 precincts, or 37 percent, tabulated more ballots than the number of voters tallied by workers in the poll books.”

How many irregular votes were counted is unknown because of obscure recount rules in Michigan, but it is clearly non-zero. Even with these problems, a recent state audit said there “is no evidence of voter fraud surrounding the presidential election in Detroit.” It was admitted, however, that “87 of the 490 [Wayne County] precinct voting machines malfunctioned”, and that the poll workers were generally old and “tired.”

FactCheck.org also missed a case in Los Angeles where the County Registrar Dean Logan was presented with “more than 80 ballots for Tuesday’s [presidential] election filled out with names and the same address.” Logan has a history with difficult elections, such as in Washington State in 2000:

The Seattle Times documented 129 felons illegally voting in that election; National Review reported nearly 350 provisional ballots were counted without being verified; and The Wall Street Journal noted some 55,000 optical-scan ballots were “enhanced” so the voters’ supposed “intent” could be determined.

There are many other (easy-to-discover yet oddly ignored) instances like these, which proves one thing: that fraudulent or improper votes have been cast in past presidential elections, and that therefore it is rational to conclude fraudulent or improper votes have been cast in this most recent presidential election. The question then becomes how many votes are illicit.

Kinds of Bad Votes

Before investigating a contentious academic study of electoral fraud from non-citizen voting, it helps to list the main sources of fraudulent or improper votes in Presidential elections:

• Legal non-citizens; i.e. those who are here legally who are ineligible to vote but do anyway
• Law-breaking non-citizens; i.e. those who have broken laws to come here and are not in the formal immigration pipeline and who vote illegally
• The dead, including legal and illegal (see below)
• The fictional; i.e. names which are entirely made up
• The multiple; i.e. citizens who vote more than once
• Felons; i.e. citizens barred from voting

The dead requires clarification. Some citizens vote early and legally and then die before the official election date. Searches afterwards might turn some of these folks up as “dead voters.” The culprit is early voting and not fraud or ill intent. Of course, names of the deceased can also be, and have been, used by the unscrupulous.

Direction manipulation, by ballot box stuffing and, if it were possible, by hacking, would largely fit under the fictional category. The Chicago example of running the same ballot through the counting machine, and the example of malfunctioning machines fit here.

In order to come to total illicit votes, estimates are needed from each source.

The Dead and Felons

The dead whose names have been used improperly do not appear to account for a large number of bad votes. Many dead people are registered, as Pew reported, but their names have not been discovered to have been systematically misused. Still, there is substance (and here) to the many Chicago jokes like this: “My father voted Republican all his life. Since he died he votes Democrat.” No one therefore knows the best estimate of dead voters, but the Pew study does give an upper bound.

Laws vary by state whether felons can vote, with most states saying convicted felons are ineligible in some way. One estimate is that just over 6 million citizens are thus restricted. Ballotpedia cites a 2008 study which discovered “33,000 convicted felons who should not be eligible to vote” in Florida; another reports the “Wisconsin Government Accountability Board announced in September 2009 that it had identified up to 195 felons who may have illegally voted in the November 2008 presidential election.”

The total improper number of votes from felons is anybody’s guess, however. That 6 million is another upper bound, with the actual total surely far less than this.

Multiple and Fictional Votes

Pew again gives a clue about multiple voting by citizens; i.e., that 2.75 million people have multiple registrations. And then Alan Schulkin, Commissioner of the Board of Elections in New York City, was filmed by Project Veritas admitting people are “bused around” to vote multiple times, predominately in Democrat-heavy neighborhoods. In 2014, North Carolina identified “hundreds of cases of potential voter fraud,” many of which were likely multiple voters.

Fictional and incorrect names can be “on paper,” as when fraudulent registrations are entered, or virtual, as with ballot stuffing and hacking. For an on-paper example, recall the infamous Acorn voter registration drives in which the group turned in “‘massive numbers’ of duplicate registration cards,” cards for fictional characters and children and others with forged signatures. How many of these faked registrations turned into real votes nobody appears to know.

Ballotpedia quotes from a National Review article in which “undercover agents with New York City Department of Investigations ‘showed up at 63 polling places [in the fall of 2013] and pretended to be voters who should have been turned away by election officials … in 61 instances, or 97 percent of the time, the testers were allowed to vote.’” Other examples exist.

Ballot Stuffing and Hacking

The Chicago example suffices for a ballot-stuffing example, though many say newer voting machines make the practice difficult. Yet computerized machines open the possibility of hacking. The Detroit example with more votes than voters also falls into this category, regardless whether the intent was malicious or due to faulty equipment.

Hacking is more mysterious. A security researcher demonstrated to Forbes the ease which some machines can be tampered with. Others agree. There were claims from some that machines in Texas changed votes from Trump to Hillary, though some of these curious changes could be put down to voter error. But the same mysterious switches from Trump to Hillary were reported in Pennsylvania and other places, too.

“I went back, pressed Trump again. Three times I did this, so then I called one of the women that were working the polls over. And she said you must be doing it wrong. She did it three times and it defaulted to Hillary every time,” Bobbie Lee Hawranko told KDKA.

The key line to the story: “Officials recalibrated the machines and said the issue has been resolved.” Here is a video of an instance of switching.

And, as all know, there were multiple reports that Russia hacked the election, but here it is generally meant that “entities” in Russia provided the emails from the DNC and John Podesta, which is a different kind of thing.

As above, the conclusion is that nobody knows the size of the error or fraud from these categories, except to say that it is not zero. The direction of reports of vote switching is, of course, interesting.

Non-Citizen Voting

In 2014, Jesse Richman, Gulshan Chattha, and David Earnest published “Do non-citizens vote in U.S. elections?” in the journal Electoral Studies. Their study relied on data provided by the Cooperative Congressional Election Study (CCES), which is an on-going Internet survey.

The CCES includes a question asking whether the respondent is a US citizen, and others asking whether the respondent voted and for whom. Some respondents who said they were non-citizens also said they voted, which of course is illegal. In the 2008 election, Richman and Earnest calculated that “more than 80 percent” of non-citizens who illegally vote did so for Obama. So pronounced was the Democrat tilt, “we find that this participation was large enough to plausibly account for Democratic victories in a few close elections.”

If extrapolated to 2016, these findings support Mr. Trump’s claim. But Richman’s paper was not well received by Democrats and the mainstream press. How accurate are Richman’s findings?

Not so accurate, say Stephen Ansolabehere, Samantha Luks, and Brian Schaffner, who penned the rebuttal paper “The perils of cherry picking low frequency events in large sample surveys” in Electoral Studies. The gist of this article is that if there is a known error rate in answering the question on citizenship, then, given the observed data, it is possible all the people who responded they were non-citizen voters were in error, and that, in fact, no non-citizens voted.

Yet there is no known error rate, only a guess, a guess which is disputed in turn by Richman (also here). Ansolabehere and the others also fail to consider what errors in answering other questions might mean. And both Richman and Ansolabehere fail to consider the biggest source of uncertainty, which is lying.

Some non-citizens vote because they honestly believe they are allowed to, but others vote knowing of the illegality. Both categories of votes are, however, illegal. A key point of dispute in the CCES is that some people who answered they were citizens in 2010 later said in 2012 that they were non-citizens (and vice versa). This supports the measurement error theory of Ansolabehere. But it also supports the theory that some might have been lying in 2010 and later changed their mind. Even stronger, there were a very large number of folks who said they were non-citizens consistently, and Ansolabehere’s approach would be to toss all these out, a move for which he does not have a solid justification.

How many who were non-citizens who claimed to be citizens, i.e. how many lied, is not known by anybody. This was an Internet survey and people were tracked through time. It is unclear how much trust respondents had in the privacy of their data; plus, the motivation to lie about voting illegally is obvious enough.

Estimating Non-Citizen Votes

Many in the press, and even Ansolabehere, intimate Richman did not consider the effects of measurement error (of mistakenly answering the citizenship question), but this is simply false. In the original paper, Richman gathered as much evidence as they could to support their claim of non-citizen voting. Most of this evidence was indirect, as in comparing demographic and other characteristics of non-citizen voters and non-citizen non-voters. But, really, this is all that can be done short of tracking down the original respondents and investigating (not questioning) them individually.

Supposing Richman is correct, his estimate of some 834,000 votes cast illegally in 2016, and most of these for Hillary, is too precise. That number has substantial uncertainty, even accepting Richman’s analysis. And then we have to add the uncertainty due to the survey itself: how were people gathered, what biases it has, how many lied, and so forth. And even if we could do all that, the result is not observational proof of the number of illegal non-citizen votes. It will be just the number (or a range, really) from some statistical model, which would be disputed until Kingdom come. Richman himself agrees more would be needed.

Lastly, another difficulty is that the CCES data does not distinguish between legal non-citizens and those who broke the law to come here. Since the CCES was an Internet survey, and thus would require access to some kind of (expensive) device, it may be that it is biased toward legal non-citizens. Folks who break the law to come here to work washing dishes or picking crops aren’t, one surmises, as likely to participate in surveys. How likely this latter group is to vote (illegally and encouraged or bused by men like Schulkin mentioned) is unknown.

Final Numbers

It is disappointing, but the answer at this date is that there is no answer, no precise answer. There is more than sufficient evidence to confirm that some illegal votes were counted, however. Much of this evidence is circumstantial, but it is also substantial.

We should cheer Mr. Trump’s call for a thorough investigation, but we must acknowledge his estimate of illicit votes was seat-of-the-pants — which doesn’t mean wrong. A serious investigation is the only way to know.

There were about 130 million votes cast in 2016. In order for Trump to have won the popular vote, about 2% of these would have to be fraudulent in Hillary’s favor. Given the possibilities listed above, it is not impossible that this was so, but the low bar for fraud is also not proof that it happened.

We should therefore cheer Mr. Trump’s call for a thorough investigation, but we must acknowledge his estimate of illicit votes was seat-of-the-pants — which doesn’t mean wrong. A serious investigation is the only way to know.

To support this site and its wholly independent host using credit card or PayPal (in any amount) click here