The Presidential Commission on Election Administration (PCEA, or Bauer-Ginsberg) released its report this week. While I support nearly all the recommendations in the report, they miss a key point about data modeling.
First, the good news: The report provides excellent recommendations on electronic voter registration, early voting, absentee ballot tracking, the use of school as polling places, and the availability of resource calculators (this tool is better than mine). And I was very happy how much the committee relied on the work of professors Stewart, Ansolabehere, and others.
However, I think the recommendation of local officials to monitor wait times themselves is misguided (bottom of page 43, top of page 44). Specifically, the committee’s call to action is:
Addressing long lines requires systematic procedures to spot when and where long lines occur. Keeping track of wait times at individual polling places can be done using simple management techniques, such as recording line length at regular intervals during Election Day and giving time-stamped cards to voters during the day to monitor turnout flow.
Election officials have a ton to worry about on Election Day; adding to their stress by suggesting they hand out time cards is an unrealistic and unnecessary burden. We shouldn’t expect expect local election officials to go through the trouble of exactly monitoring wait times when post-Election Day they would have results for only (say, for example) the 15 precincts in their county, many of which are likely homogenous. To produce great models, data analysts need many data points and heterogeneous input variables. Both criteria help the model understand exactly what happens under different conditions, and local election officials may not be able to satisfy either one.
A better approach would be to fund or promote national projects that inform us about our election system. For instance, Prof Stewart’s survey provided information on wait times, and Foursquare’s “I Voted” badge lets us know when people vote (middle of infographic). We should expand these existing efforts, as well as fund new ideas, such as a web app that records wait times (ala TSA wait times). These data and models can then be integrated into the next generation of resource calculators, which are easy to disseminate to local election officials. These officials are incentivized to use the calculators to double check that they have the proper amount of equipment assigned to each precinct’s polling location, ensuring they don’t have egg on their face later.
In sum, let’s not give local election officials more work; instead let’s provide them with the tools they need to conduct successful elections.