IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Data on Voter Behavior and COVID-19 Fuels Student Research

Undergraduates from Rice University worked with the Harris County, Texas, Clerk’s office to learn how the pandemic affected in-voter preferences, like mail-in and drive-through voting, and impacted election outcomes.

Closeup of a person reaching out of their car window to place an envelope in a collection box.
Shutterstock/Kristen Prahl
MetroLab Network has partnered with Government Technology to bring its readers a segment called the MetroLab Innovation of the Month Series, which highlights impactful tech, data and innovation projects underway between cities and universities. If you’d like to learn more or contact the project leads, please contact MetroLab at info@metrolabnetwork.org for more information.

In this month’s installment of the Innovation of the Month series, we highlight the winners of the MetroLab Student Cup, a competition among MetroLab member universities as part of their 2020 Summit. The winners from Rice University explore changes to voter preferences and behavior during the COVID-19 pandemic. MetroLab’s Ben Levine spoke with Mason Reece, Carolyn Daly and Arisa Sadeghpour on the background and development of the project.

Ben Levine: When and how did you recognize the need for this project?

Mason Reece: In March, as the pandemic began to set in around the country, a team of Rice University faculty and Harris County, Texas, election administrators formed a collaboration to work on effectively running a presidential election during a pandemic. Shortly after this team was formed, we developed the goal of designing, implementing and analyzing surveys to understand voters’ preferences and intentions for elections during COVID-19. We are all undergraduates at Rice University, and we had a shared goal of making this year’s election safe, fair and accessible. Our faculty mentors, Dr. Elizabeth Vann and Dr. Robert Stein, were awesome about giving us agency and letting us really shape the research project.  

Carolyn Daly: Primary elections in Georgia and Wisconsin that reported long lines, mail-in ballot delays and possible spread of COVID-19 were very concerning to us as voters, and to the county clerk, who runs elections in Harris County. The Harris County elections in July and November were going to require new solutions in order to be safe and accessible.

Arisa Sadeghpour: We have had to stay constantly up to date with every aspect of the election, following court cases, social media and the latest expert research to support the clerk. Our diverse skills and backgrounds have allowed us to dynamically adapt our research to the clerk’s present needs.

Levine: Who worked on this project, and why were you motivated to join?

Reece: I’m studying social policy analysis, political science and data science. At Rice, I’ve been deeply involved with voting-related research on a number of topics, so this project was a natural fit for my interests. I served as this year’s Election Day presiding judge for the Rice polling location and promoted student civic engagement by recruiting only student poll workers. The dual experience of researching effective methods and implementing those same methods further motivates me to make an impact on the voting process.

Daly: I’m studying social policy analysis, history and Spanish. It’s important to me to find ways to apply academic knowledge to something that will benefit Houston. I’ve long been passionate about voting rights — I’ve been part of a nonpartisan civic engagement club on campus since my freshman year. I also helped to co-found and lead the Houston Youth Voters Conference, an event that brings together all area high schools and colleges to get informed and excited about voting. This research allowed me to combine my academic and extracurricular interests to make a positive impact.

Sadeghpour: I’m studying computational and applied mathematics and civic leadership. I am interested in using data science for civic causes. I loved getting to experience how collecting and analyzing data can be used to inform and advocate for new election reforms. I came into the project with little knowledge of election administration, but I learned a lot and was a resource for my friends and family as they navigated voting this year.

Levine: What did the survey process entail, and how did you analyze the results?

Daly: Our surveys were designed to gather intentions and preferences of voters while simultaneously providing them with updated information on voting this fall. Thanks in part to the generous support of Rice University’s BRIDGE and COVID-19 Research Funds, we were able to purchase L2 political data to acquire a representative email sample of 330,000 voters in the county. Using Qualtrics software, we surveyed half of them in August, and then in September we surveyed the other half with updated questions on the clerk’s new initiatives. We also re-surveyed 1,838 respondents from the August survey. In total, we gathered 12,524 responses.

Sadeghpour: We included a randomized, controlled experiment in the survey to carefully evaluate voter preferences under varying threats from COVID-19. We also used detailed question logic to make sure voters only saw questions relevant to them. We weighted the survey results to match the county demographics using ANES ranking so our results were generalizable. For analysis, we did everything in R, writing over 3,400 lines of code, extensively cleaning it, building multinomial classifiers and visualizing our results. You can see one of our visualizations about changes in how voters will cast their ballots here:

A+visualization+created+in+R+showing+the+change+in+how+voters+expected+to+cast+their+votesin+the+2020+election+to+how+they+actually+did+so


Image 1: A visualization created in R showing the change in how voters expected to cast their votes to how they actually did so.


Reece: Throughout the research process, we continuously communicated with the County Clerk’s Office. They used our results to secure $12 million in additional funding to implement new voting initiatives. They also used our research to defend the new initiatives in court.

Levine: How did the results change between the first survey and the second? What factors influenced this change, and did you see anything surprising?

Sadeghpour: Most voters don’t change their voting method from year to year. They find something that works for them and stick to it. However, this year, with COVID-19 and a months-long national discourse on voting, many voters carefully re-evaluated their voting method.

Daly: In our first survey in August, over half of Election Day voters switched to early voting and many others switched to mail-in voting. Many early voters also switched to mail-in voting (see Image 1). Voters were avoiding health risks from voting in-person. This trend continued in September, as we saw increased numbers of voters moving to safer methods of early voting and mail-in voting. Between September and the election, even more voters shifted to safer methods.

Reece: However, in August, 16 percent of historically mail-in voters switched their intentions to voting early, which doesn’t make much sense from a voter safety perspective. We saw voter comments in our open response section hinting at fraud concerns, so in September we strove to carefully uncover their reasoning. As we expected, over 70 percent of eligible mail-in voters only somewhat trusted the mail-in ballot system or didn’t trust it at all. Moreover, nearly 70 percent of eligible mail-in voters who weren’t voting by mail this year told us their lack of confidence in the system was their main motivator for not voting by mail. We can see in these results how uncertainty around the postal service, mail-in ballot legitimacy and election security manifests itself in voter decision-making. 

Levine: How do you think the gravitation toward “safer” forms of voting in this election will influence future elections? Will we return to prior preferences after the pandemic?

Reece: The threat of COVID-19 prompted state and local governments to take action to address voters' safety concerns. In addition to a historic presidential election, we believe that these changes, including more flexibility to vote by mail, contributed to record voter turnout across the country.

Sadeghpour: In particular, Harris County broke all kinds of turnout records this year, with 68 percent of eligible voters casting their ballot, slightly higher than the national average in an already record year. That’s especially impressive given Texas’ historically low voter turnout and voter suppression.

Daly: Our findings demonstrate that voters can and do change their voting behavior when new opportunities and information become available. This year, Harris County added polling locations with drive-through voting, an option in which any voter can cast their ballot in-person without leaving their car. Many of the voters who switched to early voting and drive-through voting were from subpopulations who are historically underrepresented at the polls, including voters of color and young people. We estimate that about a quarter of voters who used drive-through voting have never voted in a Texas election before. New, innovative voting methods that the government implements to directly address voters’ needs can and will change voter behavior.

Reece: I think what’s particularly exciting is that many changes are not COVID-19 dependent. Expanded early voting locations and times, in addition to lenient mail-in voting policies, have every reason to stay in the future. I hope that voters who have experienced these improvements advocate that their election administrators make those changes the standard. 

The+Rice+University+student+team%2C+from+front+to+back%3A+Arisa+Sadeghpour%2C+Mason+Reece%2C+Carolyn+Daly


Image 2: The Rice University student team, from front to back: Arisa Sadeghpour, Mason Reece, Carolyn Daly.


Levine: What are the next steps for your project? Can this be scaled to other cities and other elections?

Reece: I definitely think that many of Harris County’s successes can and should be replicated across the country. I am personally excited about the Clerk’s new drive-through voting method, which all three of us utilized this year to cast our ballots. Based on our personal experience and the overwhelmingly positive public response, we think parking lots around the country could be effectively transformed into polling locations. As a team, supporting the county in their election administration has always been our main focus, and we are excited to meet and work with the new elections administrator, Isabel Longoria, to implement these innovations and more in future elections.

Sadeghpour: We also think our surveying methods and research approach could be easily applied to any community movement. Research may never perfectly mirror the real world, but it can help inform how to address numerous challenges and give projects a better chance for success. We plan to take these skills with us wherever we end up in the future.

Daly: More personally, we are hoping to publish several academic papers using our data. Right now we’re working on a paper on mail-in voting and how it relates to partisan dialog, and a paper on voters’ changing behaviors and intentions in response to the threat of COVID-19. I hope the formal realization of our research will help spread our knowledge and encourage other scholars to engage in similar partnerships with their election administrators. I am particularly hopeful that other university students like us will be inspired to search out their own meaningful research partnerships and make an impact.

Lauren Kinkade is the managing editor for Government Technology magazine. She has a degree in English from the University of California, Berkeley, and more than 15 years’ experience in book and magazine publishing.