Alphabet Soup and Voter Bias: The Effects of Ballot Order Randomization on Gender- and Ethnicity-Based Voter Biases
P12-1
Presented by: Rachel Bernhard
Prior work finds that candidates benefit from being listed first on the ballot. This seems to be true across countries, especially in local, nonpartisan, and complex elections (Australia: King and Leigh 2009; Britain: Webber et al. 2014, Rallings et al. 2009; Colombia: Gulzar et al. 2021; Korea: Jun and Min 2017; Spain: Lijphart and Pintor 1988; US: Meredith and Salant 2013, Koppell and Steen 2004, Miller and Krosnick 1998). Accordingly, many states and countries implemented randomly0generated alphabets to ensure no candidates systematically benefit.
A separate literature examines how cognitive biases shape election outcomes for marginalized candidates (e.g., Bauer 2020a, 2020b, Berinsky et al. 2020, Crowder-Meyer et al. 2020, Fulton and Dhima 2020). This scholarship often finds that these candidates fare worse in low-salience and low-information contexts like on-cycle local elections (Anzia and Bernhard 2021, de Benedictis-Kessner 2018). This is theorized to be the result of increased reliance on heuristics and stereotypes (McDermott 1997, 1998).
Taken together, these literatures produce a natural question: does being listed first benefit historically marginalized candidates more or less than ethnic majority and/or male candidates? We explore this possibility in two datasets. In the first, we use data on California ocal elections from 1995-2020. We test whether women and ethnic minority candidates are helped or harmed by randomly being listed first on the ballot relative to men and white candidates. We combine this with a survey experiment manipulating hypothetical candidates’ ethnicity and gender, candidate salience through random ordering, and election salience by varying the office.
A separate literature examines how cognitive biases shape election outcomes for marginalized candidates (e.g., Bauer 2020a, 2020b, Berinsky et al. 2020, Crowder-Meyer et al. 2020, Fulton and Dhima 2020). This scholarship often finds that these candidates fare worse in low-salience and low-information contexts like on-cycle local elections (Anzia and Bernhard 2021, de Benedictis-Kessner 2018). This is theorized to be the result of increased reliance on heuristics and stereotypes (McDermott 1997, 1998).
Taken together, these literatures produce a natural question: does being listed first benefit historically marginalized candidates more or less than ethnic majority and/or male candidates? We explore this possibility in two datasets. In the first, we use data on California ocal elections from 1995-2020. We test whether women and ethnic minority candidates are helped or harmed by randomly being listed first on the ballot relative to men and white candidates. We combine this with a survey experiment manipulating hypothetical candidates’ ethnicity and gender, candidate salience through random ordering, and election salience by varying the office.