For citizens and scholars alike, the quality of electoral representation in a democracy is central to its performance and its legitimacy. However, what effective electoral representation requires in a well-functioning democracy remains contested (e.g. Pitkin 1967, Mansbridge 2009, Sabl 2015, Wratil and Wolkenstein 2021). One area of debate is whether and how far descriptive representation – or the representation of citizens by legislators sharing their social identities and lived experiences – relates to substantive representation, or the representation of citizens’ preferences and interests in law-making. Another open question is how citizens prioritize between multiple overlapping social identities – e.g. social class, gender or ethnicity – if and when they demand improved descriptive representation.

With support from the Selten Seeding Grant (University of Cologne), we aim to contribute to this debate by running an original conjoint experiment that explores, first, how citizens prioritize different social identities when they demand improved descriptive representation, and second, whether these demands are motivated by their inferences about candidates’ preferences and priorities (and so, expectations of improved policy congruence) versus their preferences for equal representation.

As in some previous research, we theorize that citizens make assumptions about whether in-group candidates will share their preferences and priorities. They may then demand descriptive representation on dimensions linked to their issue priorities, if they think in-group representatives will share their preferences on key issues. However, prior research has also established a general preference for equal descriptive representation among citizens (e.g. Wäckerle 2022). This may also produce demands for improved descriptive representation, even at the cost of reduced policy congruence.

To compare the importance of these two mechanisms, we will conduct a conjoint experiment where respondents are asked to choose between two candidates with different social identities (some potentially congruent with the respondent), and espousing different policy positions. This allows us to ascertain how respondents weigh different social identities, and whether these weights derive from their expectations regarding improved policy congruence on particularly important issues, or from preferences for equal descriptive representation on some dimensions.

We intend to survey approximately 2000 British respondents, composed from two equally sized samples. The first sample will reflect the British population in terms of age, gender, ethnicity and educational background. The second sample will only include ethnic minority Britons, ensuring a sample of British citizens with at least one social identity that is descriptively under-represented. This two-fold sampling strategy ensures sufficient statistical power to study citizens’ priorities over social identities, including when these are minority identities.

Our project significantly advances on previous work on this topic by taking an intersectional perspective. Although some research has considered how representatives’ intersecting identities may impact their electoral performance (e.g. Hughes 2011, Ward 2016), most research on descriptive representation has only considered a single identity, like gender or ethnicity (e.g. Lerman and Sadin 2016). As such, we are able to answer the question of what drives citizens’ demands for descriptive representation, and how it relates to their demands for substantive representation, while taking intersectional dynamics into account.