ABSTRACT
Recent high-profile investigations into the U.S. judiciary have demonstrated that judges often refuse to recuse themselves from cases where they hold a conflict of interest. Given women’s presence in various political institutions has been shown to decrease perceptions of corruption and increase trust and legitimacy, I use a survey experiment to assess whether the public is more trusting of the decision-making of women judges who refuse to recuse themselves in cases where they hold a financial or interpersonal conflict of interest. I find women judges who refuse to recuse are trusted more to be unbiased in their decision-making, relative to male judges. Importantly, I find that men and women respondents are not equally trusting of women judges in this context. Specifically, men are more trusting of women judges than women.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
2 The Code of Conduct can be found here: https://www.supremecourt.gov/about/Code-of-Conduct-for-Justices_November_13_2023.pdf.
3 The report can be found here: https://www.supremecourt.gov/publicinfo/year-end/2021year-endreport.pdf.
4 Articles related to the investigation can be found here: https://www.reuters.com/investigates/section/usa-judges/.
5 For instance, a difference of means test analyzing sexism scores between male and female respondents in this sample shows that men holder higher levels of sexist attitudes than women do (means are 5.63 for men and 4.75 for women, p < .001).
6 Further, it is common to distinguish between various types of political scandal when evaluating politicians using observational data (Basinger Citation2013 ; Newmark, Vaughan, and Pleites-Hernandez Citation2019; Rottinghaus Citation2023; Stockemer and Praino Citation2019) as well as in survey experiments (Barnes, Beaulieu, and Saxton Citation2018; Doherty, Dowling, and Miller Citation2011, Citation2014; Funk Citation1996; Schönhage and Geys Citation2023; Wolksy Citation2023).
7 OLS models were used because results are similar to ordered logit but substantive effects are easier to interpret (Angrist and Pischke Citation2009). Ordered logit models are included in section F of the online appendix as a robustness check.
8 Details in section B of the online appendix.
9 Given the experimental nature of my data, I do not include demographic controls here (see Lenz and Sahn Citation202Citation1; Montgomery, Nyhan, and Torres Citation2018). A balance test is included in section E of the online appendix and the models with controls for respondent gender, party identification, knowledge of the Court, sexism, ideology, education, race (white), and age are in section F of the online appendix. The substantive results remain the same.
10 Substantive results hold when including full demographic controls for respondent gender, party identification, knowledge of the Court, sexism, ideology, education, race (white), and age. Details are in F of the online appendix.
11 The results hold when using an ordered logit model. Details are in section G of the appendix.
12 One might be curious about the role of respondent sexism in evaluating male and female judges. I ran a similar model interacting judge gender with respondent sexism and did not find an effect.