Why I don’t believe there were 3 million Illegal votes in 2016 — and you shouldn’t either

Did 3 million non-citizens really vote in the 2016 presidential election, as feverish headlines on Breitbart and Alex Jones’ InfoWars — not to mention the Tweeter in Chief himself — have proclaimed? Or was this just more fake news in a season infested with it?

My journey down this rabbit hole started on social media, as most things seem to do these days. Sister #2, the lone Trump supporter in our family clan, asked whether all the “illegals” who voted in the last election would be punished.

What illegals are you talking about? I asked, starting yet another Facebook conversation that would not end well.

Eventually my search for the source of this dubious news lead me to Twitter, where I found this, and responded thusly:

Since then, this tweet exchange has found its way to Snopes, The Inquisitor, and the Washington Post’s Fact Checker column. It’s gotten me a rash of angry responses from Trumpetizens (“fuck off” being a typical one); a tsunami of links to YouTube videos and other sources ‘documenting’ false voter registrations, some more dubious than others; and a few ‘attaboys’ from the other side.

But it hasn’t gotten me any closer to an answer. And Gregg Phillips, the man behind the original tweet and the hubbub that followed, has so far declined my invitations to discuss his data. (He also blocked me on Twitter.)

“Do your own research!” more than one angry tweeter user shouted at me. So I did.

Dead man voting

First, let’s look at the claims.

To be able to confidently state that 3 million ‘non-citizens’ voted in the US, you would need, at a minimum:

1. A complete or near-complete record of everyone who voted in the most recent election; and

2. A way to compare that record to databases of registered voters who are dead (but still showed up at the polls), people who live in the US but are not citizens, felons who live in a state that denies them the right to vote, and illegal immigrants.

You can acquire the public registration information for all 190 million registered voters in the US, but it’s expensive. NationBuilder, for example, will share its national database of registered voters for a mere $10,000 a month with a minimum one-year contract, or $120,000 total. Or you can go to each state’s registrar and collect it manually.

Most states limit who can legally access that data — usually just people running a political campaign, fund raising for issues, or academics, notes Emily Schwartz, vice president of NationBuilder.

But that database will take months to update with information from the most recent election. Right now, we only know the identities of those who voted early, says Laura Quinn, CEO of Catalist, another vendor of voter registration information that works largely with not-for-profit organizations.

“That’s about 40 percent of the ballots cast,” she says. “As for ballots cast on election day, the first two or three states won’t be reporting until December, and we probably won’t know all of them until next April.”

(A similar national voter database leaked on the Web last December. But that one won’t contain any information made available after its release — like who voted in the last election.)

Even if we had sufficient data to determine who voted, there’s a second problem: Determining which of them voted illegally.

In 2012, Pew Research revealed that 1.8 million US citizens registered to vote are actually dead. (It appears dead people have the rude habit of not updating their records or leaving a forwarding address.) Another 2.75 million moved or are serving overseas, and thus are registered in more than one place. And some 12 million records contain incorrect addresses.

(The primary author of that report, David Becker, has since stated that voter registration rolls are much cleaner now than they were four years ago when he did his study.)

Again, though, the question is: How many of these dead people voted in 2016? How many people registered in more than one place actually voted in more than one place?

To find dead voters, you would have to be sure the person who voted was impersonating someone listed as deceased by the Social Security administration. That would require being able to positively identify them, usually via their Social Security Number, to avoid mistaking them for live people with the same name. While 27 states collect all or part of a voter’s SSN (or a state issued ID number) when they register, the states don’t share this information publicly, because that would be a violation of voter privacy.

(Also: The Social Security Death Master File — yes, that really exists — contains the names of every American for whom death benefits were issued. But it is incomplete, expensive, and must be searched one name and SSN at a time.)

Matching illegal immigrants to the names of actual voters is even harder. There is no central data clearinghouse listing undocumented persons. The Systematic Alien Verification for Entitlements (SAVE) database, which can verify a person’s immigration status, is designed to determine their eligibility for government benefits. And while some states (like Florida) are trying to use SAVE to purge non-citizens from voting rolls, they really shouldn’t, says the American Council on Immigration:

“…the SAVE Program was not designed to verify whether an individual is eligible to vote, and using SAVE for this purpose will likely lead to denying U.S. citizens the right to vote…. It is NOT a database or list of all non-citizens. There is no national database of all citizens that states can check to prove U.S. citizenship or voter eligibility.”

Of course, if you are in this country illegally, odds are you haven’t registered with SAVE. It’s hard to prove a negative, but none of the sources I contacted for this story were aware of a database of undocumented aliens.

So, to recap:

1. We don’t yet know who voted in the last presidential election, and won’t for several months;

2. We don’t know if any of them were dead or unqualified to vote because they lacked citizenship status.

So where does this “verified” “3 million non-citizen voters” statistic come from?

Short answer: We don’t know. And Phillips isn’t talking.

But he might be referencing a study often cited by vote fraud activists that estimated the number of non-citizen voters anywhere from 38,000 to 2.8 million — a pretty dramatic spread.

Shortly after it was published, that study’s methodology was eviscerated by the Harvard researchers who actually collect and manage the data it was based on. The problem: The sample size was too small (less than 900 self-reported non-citizen voters) and prone to “large prediction errors” when applied to much larger groups.

A ‘privatization fiasco’

Now, let’s look at the person who’s making these claims.

Gregg Phillips is founder and CEO of AutoGov, a software company with addresses in Carol Stream, Illinois, and Birmingham, Alabama. (Phillips himself is based in Austin, according to his LinkedIn profile.) According to the company website, AutoGov uses data mining and predictive analytics to screen applicants for medicare eligibility, and claims to have processed more than 30 million cases.

So it makes a lot of sense that Phillips would be making noise about ‘non-citizen voters.’ He and his company could stand to profit handsomely if, say, a state government hired his firm to verify voter registrations and purge the rolls of non-citizens, dead people, ex-felons, and others prohibited from voting in that state.

Let’s hope that doesn’t happen, though, because Phillips’ track record isn’t very shiny. In 2004, he served under Governor Rick Perry as the Executive Deputy Commissioner for the Texas Health and Human Services Commission, where he lead an effort to privatize food stamp programs that failed in spectacular fashion.

USDA officials called the Texas food stamp privatization effort a “blot on the state’s performance record,” and threatened to withhold $173 million in funds from the state unless it improved how quickly it processed applications.

In March 2010, the Dallas Morning News reported that Phillips’ firm, AutoGov, had been hired to clean up the mess created by the privatization plan Phillips had helped bring about:

A former state official who played a major role in the state’s biggest privatization fiasco is now making money trying to help Texas fix the problems that resulted.

Gregg Phillips was the state’s № 2 social services official several years ago, and he led a push to hire a private company to evaluate applications for public assistance.

Now his Austin-based company, AutoGov Inc., has received $207,500 since November to help the state eliminate errors in deciding whether an applicant gets food stamps or other aid and how much recipients get. AutoGov was hired without other companies having a chance to bid for the work.

On top of all that, a January 2005 investigation by the Houston Chronicle uncovered numerous ties between Phillips and private firms that received lucrative state contracts in both Texas and Mississippi, where he was Executive Director for the state’s Department of Human Services in the mid-1990s. A 1995 report by a Mississippi Legislative committee concluded that Phillips’ financial connections to government contractors raised “the appearance of impropriety and could constitute a violation of state ethics laws.”

True vote detectives?

Phillips is also a board member of a nonprofit organization called True the Vote, which issued a statement in support of Donald Trump’s claims about illegal voting, but with the following caveat: “We are still collecting data and will be for several months, but our intent is to publish a comprehensive study on the significant impact of illegal voting in all of its many forms.”

True the Vote was the subject of a number of investigative profiles in the 2012 election season by the New York Times, The Atlantic, and Pro Publica. It has its roots in a Houston-based Tea Party group called the King Street Patriots.

True the Vote was one of the conservative organizations given extra scrutiny by the IRS in 2010 to determine whether it was engaged in direct political activities (which would disqualify it for 501c3 status). The IRS did ultimately grant TTV tax-exempt status, but the group is suing the agency anyway.

Depending on your point of view, TTV has engaged in either voter fraud prevention or voter intimidation in several local and national elections. But the efficacy of its methods has been questioned.

In 2011, TTV jumped into the effort to recall Wisconsin governor Scott Walker, using an army of volunteers to scrutinize the 1 million signatures on the recall petition to determine if they were valid voters. Per the New York Times:

“True the Vote said 63,038 were ineligible, 212,628 required further investigation and 584,489 were valid.

The [government] accountability board concluded that about 900,000 signatures were valid… and called True the Vote’s work ‘significantly less accurate, complete and reliable than the review and analysis completed by the G.A.B.’”

Among other things, TTV rejected signatures if people used initials instead of their full legal names, declared a handful of voters living in Madison and Milwaukee as “out of state,” and failed to recognize abbreviations in some street addresses.

Fun with data

Sometime this fall, TTV unveiled an Election Crimes Database on its website, aggregating a broad collection of polling irregularities from the year 2000 onward.

You can’t actually download the database — just a 60-page PDF version of it in incredibly small type — and you can’t view more than 60 records at a time on the site, which makes it difficult to analyze.

But a quick scan of the PDF reveals many duplicate entries, most of them stemming from the same handful of well-known cases. (The word “ACORN” appears 369 times in the file.) The cases appear to be mostly registration irregularities in local elections. The words “registration fraud” appear 706 times in the file, the most of any crime; “non-citizen voting” only 145 times.

The database includes people like Michelle Robinson, charged in 2001 for registering the names of 13 dead alderman for a mayoral primary in St Louis (and also for possession of a crack pipe). Her sentence: probation, community service, and transcendental meditation training.

None of the dead aldermen recorded votes, according to the local circuit attorney.

Phillips and TTV are also behind a mobile app called VoteStand that allows citizens to report “incidents” at their polling stations. Presumably, some of the data in the Elections Crimes Database came from that app.

I downloaded VoteStand and signed up using a fake name and my junk-mail-only email address. Aside from verifying my email, the app did no authentication of me personally — not even asking for a link to my Facebook or Twitter accounts, for example.

Then I began posting bogus reports. Like these:

(Independently, a friend who lives in Florida did the same thing with VoteStand.) If these false reports were flagged by the app, or otherwise noticed by TTV, there was no indication of it.

Myth America

The problem with voter fraud is not that it never happens, it’s that it’s extremely rare. Hauling buses filled with ‘dead’ voters or illegal immigrants from precinct to precinct is time consuming and impractical. It’s unlikely to have a dramatic effect on overall vote totals in a general election, and very likely to get noticed.

That’s why several academic and government studies have concluded that the notion of widespread voter fraud is largely a myth, kept alive by people with a vested interest in limiting the number of eligible voters, particularly in districts with a large minority population.

A look at TTV’s Election Crimes database essentially confirms this: The “crimes” are nearly all limited to local elections — individual bits of corruption you’ll find on both sides of the aisle in small town America. Like Robert Madon, who in 2006 paid people $20 to vote him in as mayor of Pineville, Kentucky (population 1,732).

While it is still possible that Phillips and True the Vote will somehow present compelling evidence that millions of illegal votes were cast in the 2016 election, I think it is extremely unlikely.

In the two weeks since that initial Tweet, they have gone from “we have verified the data” to “we are still collecting data and will need several months to analyze it.” By which time I am sure they hope everyone will have forgotten about it.

These people can’t even clean up their own databases or verify the authenticity of the people they’re relying on to ‘police’ the vote; why should anyone trust them to do the same for America?