Late last week, Rick linked to a story about potential double voters in Maryland and Virginia. Two groups (Election Integrity Maryland and the Virginia Voters Alliance) claimed to have discovered “tens of thousands” of electors on the rolls in both states, and hundreds of double voters. Those lists were apparently compiled by “comparing 3.4 million current voter records in Maryland and 5.5 million in Virginia” by name and date of birth. Elections officials in Fairfax County followed up with a tighter scan, looking for matching SSN digits as well … and found 17 potential double voters. (Some of the 17 are likely double voters. Some are likely the product of clerical recordkeeping errors.)
I’m all for the Fairfax follow-up. But I worry about the less careful trumpeting of the “tens of thousands” of duplicates, based on a comparison of name and birthdate alone. As Michael McDonald and I have shown, when you compare millions of records with millions of other records, statistics demonstrates that you’re going to unintentionally sweep up different people with the same name and birthdate. Different people. Not duplicates. That’s just how the math works.
This enthusiasm without precision can lead to real trouble. On the same day that Rick posted his original story, Ada County, Idaho, admitted that it had wrongfully purged 750 registrations — real, eligible voters, and more than the margin of victory in several of Ada’s primary elections in May. The problem? It purged based on name and birthdate matches from an interstate comparison of millions of records. Ada’s David Miller was purged because some other David Miller had later registered in Arizona.
Cleaning up the rolls is a good thing (and upgrading the voter registration system to follow individuals as they move, facilitating cleaner rolls at lower cost, is even better). But doing it right takes care. Too much screaming in too much of a hurry just leads to avoidable mistakes.