Forecasting a special election is hard, but picking up the pieces is (comparatively) easy.
With that principle in mind, I’m looking to last spring’s special election in the 9th Assembly District for insights into the upcoming special elections, with a particular interest in the neighboring Long Island Assembly districts. This data can also test the model that I proposed to determine the voters who will show up in the race to fill George Latimer’s senate seat.
The voters that turned out in last year’s special election, which was an upset victory for Democrat Christine Pellegrino, were older and more Democratic than the electorate in the district.
It’s not particularly surprising to see that seniors made up a disproportionate share of the turnout last year, as they historically represent a large share of the vote in low turnout elections, such as primary and special elections.
Turnout for the 2016 election, when voters got to choose the president, was more reflective of the enrollment.
Democratic voters made up about 40 percent of the electorate for last year’s special election, while currently only representing about 28 percent of the electorate. This increased share is not the norm for that district, where Democratic turnout in 2016 mirrored the district enrollment.
The other big swing in 2017 was with unaffiliated voters, who make up 25 percent of the district and only accounted for 15 percent of the vote last May.
The Republican share of the vote only deviated about 3 percentage points from the current enrollment breakdown in the 2016 general election and 2017 special election. A Republican strategist previously blamed the smaller vote share in the special election on an enthusiasm gap.
Unfortunately, my proposed special election model for the upcoming 37th Senate District special election didn’t accurately define the voters from the special election in the 9th Assembly District. Less than half of the electorate in last year’s special fit the model, which assumes the special election voters will have participated in the past three off-year elections (2015, 2013 and 2011 in that case).
The upset victory in the special election could explain the deviation from my model. For an upset to occur there has to be abnormal voter turnout, which is a problem for predictive modeling.
If you consider a lack of enthusiasm among Republicans in the special election you can see where the model starts to fall apart. Only about 36 percent of the proposed Republican voters from my model participated in the special, while about56 percent of the potential Democrats voted. If the percentages were closer you would have seen a traditional Republican victory (and a more accurate model).
Here are some additional notes about the special election voters in the 9th Assembly District
- Nearly 14 percent didn’t participate in either a primary or general election later in the year
- Nearly 97 percent voted in the 2016 presidential election
- For 78 voters this was the only time they ever voted
- About 85 percent voted in the general election in the fall
- About 61 percent voted in the general elections in 2017 and 2015
- About 68 percent voted at least once in a primary election