Democracy Fund’s Natalie Adona: “How I Learned to Stop Worrying and Love Queueing Theory”

NYC.lines.jpg

[Image via nydailynews]

As 2016 kicks into gear and election officials continue their preparations for multiple Election Days over the next 12 months, there is growing attention to techniques that can be used to plan for and manage dramatically-increased turnout at the polls. The Democracy Fund’s Natalie Adona had a recent blog piece (“Long Lines: How I learned to stop worrying and love queueing theory”) that takes a much deeper dive into line management tools that can and should be one of election planners’ best friends in the coming year:

After seeing reports of would-be voters waiting for hours in long lines to cast a ballot, in 2012 President Obama called for a new effort to improve the quality of voting in the United States. As we’ve identified in prior posts, the Presidential Commission on Election Administration (PCEA) helped renew the growing election science movement. Election science can be tremendously helpful to hard working local election officials, who must serve the needs of voters with limited resources. As it turns out, improper allocation of available resources negatively impacts the ability to keep long lines from forming. Local officials need the capacity to anticipate long lines before Election Day and, in turn, improve the voter experience.

Fortunately, the Voting Technology Project (VTP), a collaboration between the California Institute of Technology and the Massachusetts Institute of Technology (a Democracy Fund grantee), has built free, easy-to-use tools that can help election administrators run elections more smoothly, serve voters, and save time and money with little effort. Adam Ambrogi and Paul DeGregorio briefly mentioned these tools in a recent Democracy Fund blog post. Here, I’d like to take a more detailed look at the VTP election management tools and show how they’re helpful for election officials.

After extensive research and testing in the field during the 2014 election cycle, the VTP discovered (unsurprisingly) that long lines form when arrivals at the polling place outpace available resources—in other words, lots of voters coming in at once and not enough capacity to process them in a timely way. This is based on the concept of “queueing theory” or the study of how lines form. Featured in a recent report written by VTP co-director Charles Stewart, the tools developed by Mark Pelczarski, Stephen Graves, and Rong Yuan help optimize the use of resources at polling places and have the potential to significantly mitigate the impact of long lines on the voter experience.

Did I mention they’re free and easy-to-use?

The Graves and Yuan tool analyzes and makes recommendations on the data points that many election officials may already be collecting, including:

  • Arrival rates
  • Average time for check-in at registration table
  • Number of check-in stations
  • Maximum wait (the PCEA strongly recommended voters wait no longer than 30 minutes to vote)
  • Percentage of people who will be served within the maximum wait time (“Service level”)

Here’s how it works (follow along at home): Plug in your data using the “add precinct” icon at the top right of the screen. In the example provided here and mimicking the example provided in the VTP report, (fictitious) Precinct #0001 has 115 voters arriving per hour (calculated by assuming that a precinct will expect 1500 voters over a 13-hour period that the polls are open), there’s an average of 30 seconds to check-in, one check-in station, and 95 percent of people will wait a maximum of 30 minutes. Then, click “calculate.”

As you can see, the tool provides an analysis and a recommendation. Here, voters in Precinct #0001 will wait in line about 11 minutes on average and almost 8 percent will wait longer than 30 minutes. To meet the 95 percent service level, the tool recommends adding a check-in station.

Similarly, the Pelczarski tool analyzes “what if” scenarios, based in part on anticipated peak hours and other data points. In addition to the data identified earlier, you will also need to know:

  • Number of expected voters
  • Number of voting stations (aka, booths)
  • Average minutes to actually vote (for the average voter—not the ideal voter)

It’s also helpful to know the arrival pattern and other information, which you can see at the bottom of the screen. I created a comma separated values (.CSV) file using fictitious precinct and county data, uploaded it (at the top of the screen where it says “Load Precinct Data”), and toggled the arrival pattern to “Early morning peak” and got the following results:

As you can see, based on the data I provided in my .CSV file, the tool tracks average wait times throughout the day in this precinct in County #001. That precinct will experience wait times of over 30 minutes starting at about 9:30 am and will dwindle as the day progresses (at 3:30 pm, for example, the wait is only 6 minutes). I substantially reduced the wait time after I asked the tool to project how many folks might walk off—here, 28 people will potentially turn away from the polls at this precinct. Based on this data, I might have to consider adding another poll worker, e-poll book, or otherwise re-evaluate the average minutes spent at the check-in table.

The ability to potentially project the number of people who will leave the line is an incredibly important predictive tool.

Gathering all this data can be time-consuming, but this is an investment that will pay off in the short- and long-term. For some election officials, adding another poll worker to the check-in table or adding an e-poll book requires money that doesn’t exist. These tools have the potential to help officials make the case for increased funding by using hard data to justify budget increases. [emphasis added]

Anyone who hasn’t tried out these tools really should do so; even if you don’t have access to real data on arrival times, simply using best guesstimates still gives you a deeper appreciation of the impact of queuing theory on real-life lines. If you do have real data, then you have a powerful tool to be ready when the polls open this year. Either way, it’s a powerful incentive to get a firmer grip on how many voters are showing up to cast ballots and when.

This is an incredibly valuable and well-written piece that should be required reading for election officials nationwide; thanks to Charles Stewart and his VTP colleagues for their work on the tools and the math that supports them, to the PCEA (again!) for sharing them with the election community – and especially to Natalie and the Democracy Fund for making them so accessible in this post.

It should be an exciting year with typically high presidential turnouts – but at least with these tools, election officials have a chance to be forewarned *and* forearmed and avoid (or at least plan for) longer lines. Stay tuned …

Leave a comment

Your email address will not be published.


*


University of Minnesota Wordmark