Invisibility or Inclusion? Ethnic Parties, Ethnic Seats, and Gender Quotas and the Representation of Minoritized Women

ABSTRACTThis article seeks to understand the circumstances under which minoritized women are descriptively represented. Drawing from a unique dataset of 7,978 legislators in 37 countries, we conduct the first cross-national examination of minoritized women’s representation at the level of individual legislators. We find that gender quotas, ethnic parties, and ethnic seats are effective at enhancing minoritized women’s political representation across different electoral systems, especially when clustered together. And, although ethnic parties and ethnic seats promote the representation of both minoritized women and men, ethnic seats provide a more level playing field between minoritized women and men than ethnic parties.KEYWORDS: Minority womenelectoral systemethnic seatsethnic partiesgender quotas Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1. Preferred concepts and terminology for marginalized groups vary widely across countries and academic fields. In this study we use the term “minority” to describe groups that have experienced social, economic, or political marginalization, either by law or by custom, and numerically small groups that may be marginalized simply due to their size. We do not consider small groups that are socially and economically dominant to be “minorities.” We use “minority” and “minoritized” interchangeably; the latter emphasizes the social construction of majority-minority boundaries and the systematic oppression that “otherizes” group members. Like other cross-national research on policies targeting minority groups (e.g., Bird Citation2014; Hughes Citation2016; Reynolds Citation2005; Tan and Preece Citation2021), we use the terminology of “ethnicity” as an umbrella that includes groups differentiated by race, religion, nationality, language, including tribes and castes (Chandra Citation2005).2. Quotas are formal rules that guarantee that a certain percentage of candidates or elected representatives are members of a targeted group, including women and ethnic minorities (e.g., Hughes et al. Citation2019; Tan and Preece Citation2022).3. Legislator data are augmented from a dataset created by Melanie Hughes (see Hughes Citation2011, Citation2013). Although the information is outdated, more recent cross-national data on the gender and minoritized status of legislators are not currently available.4. Only one transgender legislator was among the nearly 8,000 persons in our data. Although she uses feminine pronouns, she has stated publicly that she sees herself as neither a woman nor a man, so we do not code her into either category.5. For additional details on Hughes’ coding of ethnic minorities, see Hughes (Citation2013).6. For example, treating Burundi’s Tutsis as a “minority group” is questionable. Tutsis held economic, political, and military power in the decades after Burundi’s independence. Since democratization in 1993, however, political power has rested squarely in Hutu hands.7. Research in the United States, for example, demonstrates that experiences among women of color differ significantly among Asian, Latina, and Black women (Brown Citation2014; Matos, Greene, and Sanbonmatsu Citation2021).8. Enforcement mechanisms are also often used to classify gender quotas. However, research demonstrates that placement mandates have a larger marginal impact on the election of women than enforcement mechanisms (Paxton and Hughes Citation2015; Schwindt-Bayer Citation2009).9. Since we are analyzing representation at the individual legislator level, this reduces the differences between national and party quotas to elements such as placement mandates and the percentage of the quota. While a voluntary, party-level gender quota may be considered a weak quota at the national-level because it is only applied to one party, in our dataset, it is considered a strong gender quota at the individual level as long as it includes a placement mandate. The legislator elected under a strong party quota would presumably benefit from the quota, but a legislator in the same country elected from another party without a quota would not.10. We code all districts in the U.S. that have a majority population that is less than 50% as a minority-majority district.11. Many studies do not include districts within ethnically defined regions as ethnic seats. However, as Appendix B shows, these districts clearly provide additional opportunities for the election of minorities, including minority women, much like reserved seats or minority-majority districts. In our data set all three types of ethnic seats (reserved seats, minority majority districts, and districts in ethnic federal units) elected minority women at much higher levels (ranging from 18% to 24%) than the average for all seats (2.8%). In auxiliary analyses we excluded ethnic federal districts from the data, and the substantive conclusions remain the same.12. Ideally an examination of minority women’s electoral fortunes would include individual-level data on all candidates. This would allow us to look at otherwise similar minority women candidates – some who won and others who lost – who ran for office under different institutional arrangements. However, we could not collect data on all candidates, nor do we know of any cross-national dataset that includes detailed data on ethnicity and sex for all candidates in competitive elections.13. The concern we address here is twofold: first, these groups may be too different from the other minoritized groups in our study to justify their inclusion; second, the political systems that enable their representation may not be best understood as a combination of ethnic seats and parties.14. There is no consensus on the number of level-two units required for accurate multi-level models, but one estimate suggests a minimum of 40 units are needed for random intercept models and 80 units are needed for estimating cross-level interactions. Given we only have 37 countries, we only ran multi-level models as a robustness test.15. We were concerned that small-sample bias of maximum likelihood estimation could undermine our logistic regression models (Allison Citation2012). Because Firth logit in Stata does not accommodate clustered standard errors, we prefer to use the ML estimates as our main results.16. Hughes (Citation2013) finds a similar pattern based on a dataset with over 80 country cases.17. In an auxiliary model, we also interacted PR with ethnic parties, ethnic seats, and gender quotas, and the main effect of PR demonstrated that PR systems do not benefit minority women absent these mechanisms. Results available upon request.18. See Matland and Studlar (Citation1996); as an exception, see Cowell-Meyers’s (Citation2014) work on movement-parties.Additional informationFundingThis work was supported by the National Science Foundation [SES-0703418].Notes on contributorsStephanie HolmstenDr. Stephanie Seidel Holmsten is an associate professor of instruction at the University of Texas at Austin with a joint appointment in the Government Department and International Relations and Global Studies (IRG) program. She is also associate director of IRG and co-director of the Brumley NextGen Scholars program. Her research focuses on the election of women, ethnic minorities and minoritized women. She is faculty director of the global virtual exchange learning community and leads study abroad to Chile and Paris. You can hear her on The Other Side of Campus, showcasing the teaching and research interests of UT faculty.Melanie M. HughesMelanie M. Hughes is Professor of Sociology and Co-Director of the Gender Inequality Research Lab (GIRL) at the University of Pittsburgh. Dr. Hughes is particularly interested in the ways that gender intersects with other forms of marginalization to affect women’s political power. Dr. Hughes is co-author of Women, Politics, and Power: A Global Perspective, now in its 4th edition (Paxton, Melanie & Tiffany 2020). Since 2015, she has also collaborated with the United Nations Development Program to expand the availability and quality of data on gender equality in public administration.Robert MoserRobert G. Moser is a Professor in the Department of Government at the University of Texas at Austin. He is the author of two books: Unexpected Outcomes: Electoral Systems, Political Parties, and Representation in Russia and Electoral Systems and Political Context: How the Effects of Rules Vary Across New and Established Democracies and has co-edited three books: Russian Politics: Challenges of Democratization, Ethnic Politics after Communism, and Is Democracy Exportable?. His articles have appeared in World Politics, Comparative Political Studies, Comparative Politics, Perspectives on Politics, Electoral Studies, and Post-Soviet Affairs. He served as Chair of the Government department from 2013 to 2019.