Filling in the Record Book: Election Data Analysis Can't Start Without Election Data Collection

[Image courtesy of FiveThirtyEight]

Yesterday, Nate Silver’s new and expanded FiveThirtyEight had a fascinating story on Dick Pfander,”The Man Who Preserved Decades of NBA History“, whose hobby of collecting and tabulating years of NBA boxscores (the image above is handwritten career stats for Kareem Abdul-Jabbar) formed the basis for the league’s early statistical records. Now, that archive is powering a new generation of highly-sophisticated analysis. I highly recommend the article as an example of how something that seems obvious and straightforward to collect and analyze – sports statistics – is anything but.

To an election geek, the article is timely because of a discussion that took place yesterday at the Senate Rules Committee hearing on “Collection, Analysis and Use of Data: A Measured Approach to Improving Election Administration.” Justin Riemer, former Deputy at the Virginia State Board of Elections and editor/co-author of the Republican National Lawyers’ Association response to the PCEA report, made the observation during the question-and-answer period that while he supports the robust data analysis of election administration data, he worries that the field lacks the technology and the training to actually collect the data. In particular, he said he worried about the pressure on poll workers and part-time election administrators to set up and carry out data collection procedures that might hinder their other duties.

This is not a trivial concern. While some of the current uneven state of election data is a function of a lack of awareness or commitment to data collection and analysis (what I call a “won’t” problem), there is also a significant “can’t” problem – a lack of capacity of would-be data collectors to generate the figures that are necessary for robust and complete data analysis. The challenge is to identify those people who are willing and able to be the Dick Pfanders of election data – and, eventually, giving them the tools they need to assemble and tabulate the figures more easily, as with current electronic scorebooks that do the job in basketball and many other sports.

You will be hard-pressed to find many people who disagree with the assessment that better analysis of election data is fundamental to continued improvement in the field. What we need now is a realistic approach to making sure that the data is being collected. While it may not require Dick Pfander’s fountain pens and neat handwriting, it will require his passion – almost obsession – to making sure those numbers see the light of day.