Massachusetts first participated in Crosscheck in 2014. As of November, 2017, Massachusetts announced they will no longer participate in Crosscheck stating, "it has proven to be a lot of work for very little reward".

As a result of a Freedom of Information Act request, Massachusetts shared some interesting statistics from their final data exchange in January 2017 which illustrate the data quality problems of the program.

Vital Statistics:

  • Joined: 2014
  • Quit: November, 2017
  • Not in ERIC
  • Voter Records sent in 2017: 4,491,777
  • In-State Voter Records in Result file: 191,734

Duplicate Registrants

Of the 191,734 Massachusetts voter records returned by Crosscheck in 2017, there were 181,081 unique Massachusetts voters identified as "potential duplicate registrants" in other states. Keep in mind:

  • Crosscheck only matches on first name, last name and date of birth to identify "potential matches", ignoring data such as middle name or the last four digits of social security number (SSN4) that could be used to identify true matches.
  • We're talking about potential duplicate registrants not "double voters". These are simply people who moved to another state and registered to vote.

What happens if we look at voters who are truly likely matches, by looking at SSN4? The potential matches drop from 181,081 to 22,622. That means over 87% of the data Massachusetts received were very likely not actual matches, but instead of people who share common names and birth dates.

We've seen no evidence and are not implying that the 181,081 potential matches were inactivated or purged from the voting rolls. This is simply a statement on the quality of data returned by Crosscheck, which puts the burden on state and local officials to filter out the 87% of meaningless data to find the useful data.

Double Voters

Let's look at the issue of "double voters". These are people who vote twice in the same election in two different states. Every fact-based analysis of this type of voter fraud, even those conducted by Republican officials hoping to prove that this is a widespread issue, has found that it is exceedingly rare. Generally between 0.001% and 0.002% of the total vote.

Based on Crosscheck data, Massachusetts found that 15,433 voters matched a voter record in another state where a vote was cast in both Massachusetts and that other state. Whoa! That's a lot of fraud! Hold your horses. As we saw above, once you match on criteria other than simply name and birth date, the data narrows down quickly. If we filter this to only voters where the SSN4 matches we get 88 voters. A few things to note about this:

  • This exactly matches the findings in "One Person, One Vote", a Harvard, Stanford, Penn, Yale, and Microsoft study into the (non)prevalence of voter fraud. Examining Iowa's data, they found that 99.5% of Crosscheck's reports of "double votes" were easily discounted simply by matching on SSN4. 88 is 0.5% of the 15,433 voters identified by Crosscheck.
  • If all of these 88 cases were proven legitimate cases of double voting, this would account for 0.0025% of the total vote in Massachusetts. However...
  • What we've seen from examining fraud in other states is that most of these 88 cases are likely the result of clerical error and not actual fraud. In fact, the ex-Director of Elections in Kansas, Brad Bryant stated, "In a majority of cases of apparent double votes, in the end they do not turn out to be real double votes due to poll worker errors, mis-assignment of voter history, voters signing the wrong lines in poll books, etc."

Conclusion: The vast majority of the data received by Massachusetts was not useful for finding duplicate registrants and very little of the data was useful in finding voter fraud. Once again, the statistics prove the lie to Kobach and Trump's claims that "millions" of votes are cast illegally.