Judicial Selection and State Gay and Reproductive Rights Decisions

Abstract

Partisan cues, dynamic representation, and indirect accountability provide theoretical underpinnings for the influence of judicial selection and public opinion in state courts. It is unclear, however, how their effects change across different policy domains. We begin to address this gap by examining state gay and reproductive rights decisions. The effect of national public opinion is conditional on nonpartisan elections for gay rights decisions, whereas judges in retention systems are more responsive to state-level opinion. Partisan elections are the conduit for opinion in reproductive rights cases. This only partially supports the theoretical expectations, suggesting policy domain makes a difference.

Keywords:

• judicial selection
• judicial behavior
• judicial elections
• state courts
• decision-making
• marriage equality

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Though it did not uphold a state constitutional right to marriage, per se, the Vermont Supreme Court earlier held in Baker v. Vermont (1999) that the state could not deny benefits to same-sex couples that are provided to other couples.

2 While we are more broadly interested in reproductive rights, questions specifically on abortion are far more consistent and ubiquitous than questions on other reproductive rights issues, like contraception. Thus, we rely on abortion opinion in our subsequent analysis.

3 Replication files for this article, including the opinion estimation, can be found on the corresponding author’s Dataverse https://doi.org/10.7910/DVN/8B88CY.

4 We are grateful to the author for providing replication materials for his study.

5 We followed the same procedure used by Devins and Mansker (2010). It is important to note that we do not include all potential abortion decisions, as we wished to include votes related to expansions or restrictions of individual rights. For example, we do not include cases regarding the negligent death of a pregnant woman and whether the negligent party was held responsible for the death of the fetus.

6 Judges with a missing vote value (i.e., NA) either did not participate in the case, were disqualified, or recused themselves. Also, in some cases, COLR judges did not participate in a decision because, instead, a special judge provided a decision.

7 It is not uncommon that there is disagreement over how exactly each state’s selection method is categorized. Canes-Wrone, Clark, and Kelly (2014), for instance, include California as commission-retention, but Zschirnt (2016) does not. He categorized California, Maryland, and Utah as appointment states, though judges in those states do face retention elections and appointments are made with the support of an independent commission. Maryland is a good example of the difficulty in dividing states into these discreet categories. By law, the state has gubernatorial appointment, but, in practice, governors since 1970 have maintained a nominating commission via executive order. In California, the governor presents appointment options to a commission, not the other way around. Online databases, such as those maintained by the National Center for State Courts (http://www.judicialselection.us/) and Ballotpedia (https://ballotpedia.org/Judicial_selection_in_the_states) even differ in their categorizations. Because retention election is the key to judicial accountability for us, we include all three of these states as retention states, even though the original data for gay marriage opinions did not. Thus, we follow the categorization by the National Center for States Courts.

8 For docket control, we used the Court Statistics Project’s data on appeal by right and appeal by permission caseloads to determine whether courts have mandatory (1), discretionary (3), or a mix (2) forms of docket control. This results in a 1-3 ordinal scale. See http://www.courtstatistics.org/other-pages/statecourtcaseloadstatistics.aspx for case load data.

9 For our successful reproduction and robustness checks, see the Supplemental Information.

10 Given that there are judges who split their votes between pro- and anti-reproductive rights, we ran this series of three models with two slightly different data sets. The data reported her in the main results drops the divided votes and only models clear votes. The results in the supplemental information are based on a dataset where split votes appear as two identical (save for one positive (1) and one negative (0) vote) observations. The results in the supplemental information are consistent, though slightly weaker, than those presented here. The most substantive difference is in the interaction model, though the marginal effects still show a significant effect (using 95% confidence intervals) for pro-access public opinion in states with elections. Thus, we argue that the results are robust to these changes, but it is important to report the differences.