A Tale of Three Cities – Lawrence, Haverhill and Methuen: Comparing Voter Registration Lists by the Numbers

By: Kathy Runge – November, 2014

I’ve been writing programs to look for duplicate registrations on the voter registration list in Lawrence for a few years. This list can be searched using many criteria; in this article I have used three of my algorithms. These programs can’t really tell if two registrations correspond to the same person, but the ones used for this article give few false hits.

The vast majority of duplicate registrants aren’t even aware they’re listed twice; it’s due to clerical ‘error’ or an honest lapse of memory. My programs will naturally identify a few pairs of twins or other instances of people with similar names and birthdays as potential duplicates. I’ve researched many pairs of potential twins and other coincidences in Lawrence, but not in other cities. To be fair in the analysis below, I’ve given you the raw numbers and not deleted any known twins, etc., from these lists. Assuming the rate of these occurrences would be similar in all three cities, this data should provide a fair comparison of the accuracy of their voter registration lists. I have the files if anyone is skeptical.

So what kind of a job does Lawrence do in checking that people are not already registered to vote when new registrations come in? How does our city compare to other cities around us? The latest data I have for other cities are the registration lists from Haverhill and Methuen after the updating that occurred post November 2012 election. I also happen to have a copy of Lawrence’s registration list after that election. These lists should be the most accurate of the year in Massachusetts, as after five or so years of voting inactivity and not returning the annual census, inactive voters get purged from the rolls at this time. I also have the Lawrence voter registration list that should have been used for the current election. Here is a comparison of these lists.
Similar names are hyphenations and switching of first, last and middle names. This algorithm doesn’t count different spellings, etc., as similar.

**Similar birthdays are interchanged month/day, the same month and day a few years apart, and other common patterns. Equal birthdays are not considered similar by this program.

Do these numbers bother anyone besides me? Any old clerical error or honest mistake on the part of a voter can get someone registered to vote twice and then it takes a miracle or a decade to get the duplicates removed. Sloppy record keeping makes election fraud that much easier to commit and much harder to detect. I guess the good news is that Lawrence is now checking for registrations with the exact same first, middle, and last names and birthday. Or the state is checking it for them. Shouldn’t the Board of Registrars or the Secretary of the Commonwealth be concerned about the rest of these duplicates? I can easily find a few hundred more with my other programs.

If the official lists of people that actually voted in any particular election are accurate, the number of cases of election fraud by someone using a duplicate registration is low. But let’s not take too much comfort from that, as we’re all aware that many local elections are decided by less than 100 votes, and this is only one way to commit election fraud. When I examined Lawrence’s list of registrations that were credited with voting in the presidential election of 2012, I found over 20 pairs of duplicate registrations in Lawrence and zero in the other two cities. No official was interested in these results. A similar analysis of the November 2013 Lawrence election found only 2 or 3 pairs. Isn’t it amazing what results strict oversight can achieve? I’ll be analyzing Lawrence’s list of actual voters in this year’s election, stay tuned. To anyone interested in seeing how the voter registration lists in your community stack up, I’m available!