Snow votes or no votes?: Crunching the numbers on blizzard town meeting turnout

  • Voting takes place at Three Rivers School in Pembroke on Tuesday, March 14, 2017. Polls opened at 11 a.m. and many residents came during that first hour to beat the worst of the snowstorm. (ELIZABETH FRANTZ / Monitor staff) Elizabeth Frantz / Monitor file

Monitor staff
Tuesday, April 11, 2017

As lawmakers in the State House hash out the legality of snowstorm-postponed town elections, one of the most intriguing points of debate has been how rescheduling affected voter turnout.

The question has been whether more people were kept away from the polls (“disenfranchised” is the term that gets tossed around) in towns that held elections during a blinding blizzard on March 14, or in towns that made the unprecedented decision to move voting day from its originally schedule date.

The question may not affect the legal debate, but it is interesting enough that Secretary of State Bill Gardner had his office crunch 10 years’ worth of town meeting election turnout numbers to find an answer.

He argued before a House committee meeting this month that the numbers show communities did better if they didn’t postpone voting. More communities that voted on March 14 had their highest turnout of the decade than those that moved voting did, Gardner argued.

Further, he said, more communities that moved voting day had their lowest turnout of the decade.

Fair enough, but the Monitor decided to parse the department’s data a different way: We looked at decade-long averages.

The Secretary of State’s office compiled 10 years’ worth of voting turnout data for 98 towns – 59 of which voted on March 14, and 39 of which moved to a later day. (Many other towns provided voting totals that covered only some of the years, which we did not include.)

The 59 communities that voted as scheduled on March 14 had an average turnout of 480, which was 146 fewer than their average over the previous 10 years. That’s a loss of 30 percent.

The 39 communities that moved the election had an average turnout of 987, which was 178 fewer than the 10-year average. That’s a loss of 18 percent.

On an absolute basis, towns that moved the election did worse – losing 178 voters compared to their decade average compared to towns that kept the election March 14, which lost just 146 voters.

Hooray for not moving the election!

But the towns that moved the election in this data set have more voters on average than the nonmoving towns, which means the story is flipped if you look at percentage. There, the towns that moved the election did better (losing 18 percent) than those that kept it on March 14 (losing 30 percent).

Hooray for moving the election!

Which method is more accurate, comparing absolute voter turnout or percentage voter turnout?

Sorry, it’s a judgment call. There’s no hard-and-fast answer to know who was right.

But at least it’s an opportunity to use that math you learned in grade school.

(David Brooks can be reached at 369-3313 or dbrooks@cmonitor.com or on Twitter @GraniteGeek.)