Using election data beyond the results

By Soo Rin Kim

You’re a city government reporter and you’re on a deadline to write about a public hearing on the city’s zoning overhaul. You want to quote a resident who made an interesting comment at the hearing, but you forgot to get the exact spelling of his name (oops!). What do you do?

As unlikely as it sounds, voter registration data can save your life in a situation like this. If you managed to get his address, you can look up his name in the county voter registration data.

At a 2017 CAR Conference panel, Derek Willis, a news apps developer at ProPublica; Rachel Shorey of the New York Times; and Daniel Smith, a University of Florida political science professor, discussed how to utilize election data beyond election results with some examples from this year.

ProPublica used voter survey and turnout data to investigate administrative problems in the election system – including voter registration rejection rates – as part of Electionland.

Voter registration data usually includes a person's name, address, age, gender, race, party registration and even voting history. This data can usually be obtained directly from county agencies or from private data vendors that collect and standardize the data.

Voter registration data is also handy when looking up sources’ names and addresses or doing broad demographic counts. You can also dive in deeper to find more interesting anomalies, such as Republicans that live in mostly Democratic districts and districts with big swings.

Panelists said some states are particularly helpful. Florida collects additional election information such as the number of anticipated voters, staffing and types of voting machines in Election Preparation Reports. Georgia collects additional early voting and absentee voting data.

They also warned that election data is ephemeral in nature. States and localities have different retention policies and are not required to archive data beyond certain periods of time, making it difficult to compare election cycles. County-level data can also be incompatible. It took about a month and a half for Smith’s research team to collect precinct data from all 67 counties in Florida. And the data they obtained wasn't easily synchronized because every county had different reporting methods.