Ballots and Recreation
Aug 20, 2021 – I was able to download all of the Ranked Choice ballot data from the June 22, 2021 New York City Primary Election and do some analysis and simulations of the Democratic Mayoral Primary in which Eric Adams was elected (after some unfortunate glitches in the ballot tabulation by NYC officials). Here are some takeaways from my “recreation” over the last day:
(1) Managing over a million ballots is a lot harder than the measly ~22,000 ballots we might expect for our Cambridge municipal elections. In fact, in my attempt to use the tabulation software that we use here in Cambridge (ChoicePlusPro), it broke when it had loaded about 233,000 of the 1,040,485 NYC ballots – and that was after I had deleted all of the other data related to the many other elections taking place in that primary. One of these days I’ll try again using different tabulation software, but it was worth the effort even if only to learn what the upper limit of the Cambridge PR tabulation software is.
(2) Since we began doing our PR elections in 1941, the maximum number of voters in our municipal election was 40,546 (in 1949). The minimum was 13,721 (in 2007). The median over all 40 of our municipal elections is 25,888 and the mean is 26,480. In recent years the number of people voting was 17,846 in 2013, 17,959 in 2015, 22,581 in 2017, and 21,321 in 2019. There has been much speculation about what the turnout might be this year with the availability of Early Voting and Mail-In Voting with many people expecting an increase in turnout (and some campaigns planning to affect that turnout).
(3) After I broke the Cambridge tabulation software when trying to run the NYC Democratic Mayoral Primary data, I went back to the old fashioned tools to do The Count “by hand”, i.e. using Microsoft Excel to methodically do all the sorting and ballot transfers. It took some time, but my results were quite close to the results released by the NYC Board of Elections. The only difference, I believe, has to do with how NYC handles undervoted ballots (skipped preferences) and overvoted rankings (same rank for more than one candidate). In my simulations I simply ignored all missing rankings and overvotes, but I believe NYC handles overvotes more harshly. In any case, I ended up with Eric Adams at 50.44% over Kathryn Garcia at 49.56%. The official results were Adams at 50.45% and Garcia at 49.55%. I also did an extra Round to see how many votes would be credited to Adams had Garcia’s ballots been transferred, and Adams ended up being ranked on 56.38% of all ballots cast. There were 13 candidates and voters were allowed to rank up to 5 candidates.
One of the New York City mayoral candidates was Paperboy Love Prince. It was interesting to see that when Paperboy was counted out with 4,195 votes to transfer, Maya Wiley, the candidate backed by Bernie Sanders and the Revolution, got 1,726 of his ballots – nearly four times as many transfers from Paperboy than any other candidate. – Robert Winters