Team profile by members of the project team

Well-functioning democracy requires government responsiveness to the preferences of citizens across a wide range of constituencies. The health of democracy is threatened by the prevalence of unequal participation, unequal influence, and the extent to which unequal participation and influence overlap. Freedom House’s 2021 Freedom in the World Report showed that over the last decade, the United States’ democracy score declined by over 10%. Recent events, including rising election denialism, misinformation, and state-level adoption of anti-voter policies, have contributed to this decline.

In response to the declining health of U.S. democracy, we decided to design and implement a project to measure inequality in voter turnout across the states. By focusing on inequality in voter participation, we attempted to better understand state-level electoral policies and their impact on each citizen's ability to be heard at the ballot box. We measured inequality in democracy across U.S. states by analyzing disparities in voter turnout within four different demographic categories: age, ability status, race/ethnicity and educational attainment. We then compared the turnout gaps in the most and least equal states with a qualitative analysis of state-level electoral policies. 

Using data from the U.S. Census Bureau and the Bureau of Labor Statistics, we observed state-by-state inequalities in 2020 voter turnout among all four demographic categories. We found some correlation between policy indicators and 2020 turnout inequalities for three of the demographic categories: age, ability status and race/ethnicity. We generally found that policies that give voters flexibility (such as language access, no-excuse vote-by-mail, online voter registration and expansive early voting) tend to be associated with more equal voter turnout across demographics. 

To further supplement our analysis, we developed a predictive model using multilevel regression with poststratification that incorporated weighted survey data from the six most recent federal election cycles. The model roughly approximates the inequality of historic turnout trends within the aforementioned four demographic categories. The modeling identified different states at the most and least equal end of each demographic category. 

Beyond continued improvements to the logistic regression models, there are additional demographic categories (e.g., gender and household income) that could benefit from further exploration. There may be additional confounding variables that impact the observed turnout results, including each state’s voting culture and the overall competitiveness of elections in a state. Ultimately, our project rests on the concept that democracy is best served when communities vote in proportion to their population. We hope that additional research will shed more light on unequal turnout across demographics and the policies that contribute to those observed inequalities.

Measuring Democracy

Poster by David Schanzer, Asher Hildebrand, Deondra Rose, Kristin Goss, Alicia Blanco, Allyson Barkley, James Gao, Paul Finkelstein, Mac Hoeve, Clara Bonzi Teixeira, Olivia Smith, Saralyn Carcy, Yadira Paz-Martinez, Victoria Hernandez, Darshan Vijaykumar, Karam Oubari and Eric Gim