Helping to Build a Voting Time Estimator

Featured image: close up of a ballot

Everyone in the election field knows how important it is to minimize waits at the polls, and in recent years, we’ve seen big advances in using data to help predict and avoid polling place stress.

But while there’s a lot of research on overall wait times, there’s little data out there that addresses one critical piece of the puzzle: the amount of time it takes to vote a ballot.

For that reason, the Center for Technology and Civic Life is working with software developer Mark Pelczarski to build a tool that will estimate how long it will take voters to mark a ballot based on its contents. Once it’s ready, the tool will be available for free in the Election Toolkit.

As we move forward with the project, election officials have an important role to play in helping it come together.

Working toward a goal

“We know that when voters experience long wait times at the polling place, it can discourage participation,” says Whitney May, Director of Government Services at the Center. “This project is about collecting data to discover a clearer connection between ballots and voting times. Ultimately, we want to give election officials a free tool they can use to make resource allocation decisions that minimize wait times for voters.”

These are issues that have been on Mark’s mind for a long time.

A few years ago, Mark wrote and refined a software program (the Election Toolkit’s Polling Place Resource Planner) that uses data inputs to predict polling place capacity. It’s a highly useful program, but it’s dependent on a few key pieces of data — in particular, the predicted average minutes to vote.

Planner tool with inputs for number of expected voters, check-in stations, voting stations, average minutes to vote, and so on

Mark Pelczarski’s Polling Place Resource Planner tool

“Those who have used the tool can readily see how critical a difference a minute or so in that average time can be; it can easily change a smoothly-functioning polling place into one with long lines and hours-long waits,” explains Mark.

Unfortunately, he says, this highly sensitive “minutes to vote” has often been the product of guesswork instead of real data, meaning that officials end up making important decisions based on unreliable information.

Part of the reason that determining voting times has been such a stubborn challenge is needing to account for all of the possible variations in ballots. But Mark has a plan for tackling this challenge, and his work will yield incredibly valuable information for election administrators.

“We’re using computational and mathematical analysis,” Mark explains, “to correlate differences in ballots with differences in completion times. For example, how much time does each additional contested race add to the time needed to vote? How much for each referendum? Is there a measurable time difference between electronic and paper voting? These are among over a dozen variables we’re analyzing.”

Voting Time Estimator with inputs for number of contested races, uncontested races, referendums, and so on

The Voting Time Estimator in its prototype form

Mark is refining the math and programming stages of the project, but the most important step is collecting data — lots of it. This is where we need the help of election officials who care about these questions.

Collecting ballot time data

Although we have a good start with data contributed by Charles Stewart III of MIT, we need more to be able to create a reliable estimator.

“We’d like to collect as much actual timing data as possible with a wide variety of different ballots,” Mark describes. “Ideally, we’d like 50 or more voters timed with each ballot and minimally at least 30, so that we have reliable averages and ranges for the mathematical portion of our work.”

This may sound like a lot, but data collection doesn’t need to be a major undertaking. Although we’d like to have election officials oversee several timers working at several polling places, a single timer working at just one polling place could collect voting times for over 50 voters in an hour or two.

If you’re an election official who already has data on voting times, we hope you’ll volunteer to contribute to this project by sharing your data.

If you don’t have data, we’d love to have you collect some during an upcoming election, and we’re ready with technology and support to help you.

On the technology front, Mark has developed a simple tool to streamline the data collection process.

Using the Voting Timer App he created, you can time 4 booths simultaneously, and it automatically uploads your results, so you don’t need to do any data entry. It’s a lot easier than using a stopwatch and a pencil.

The Voting Timer App, with 3 booths actively being timed and a total average time of a little over 3 minutes for 14 voters

The Voting Timer App

In terms of support, CTCL is offering training and assistance at no cost.

“If you’re interested in collecting ballot time data,” Whitney explains, “CTCL can train you and your team to use the Voting Timer App to collect voting times at your polling places. We want to make it as easy as possible for election officials to contribute data, so all that’ll be required from your office is a few staff members and their time.”

We hope you’ll consider contributing to this project. By helping us build this tool, you’ll also be helping the countless election officials around the country who will benefit from it.

If you have voting time data that you can share, or if you’d like to talk with us about collecting data in your area, please get in touch by emailing Kurt Sampsel at kurt@techandciviclife.org.