Explaining the nomination of ethnic minority candidates: how party-level factors and district-level factors interact

ABSTRACT

In this paper, we explain the nomination of ethnic minority candidates for lower house elections. We argue that these nominations are explained by the incentives that different parties face in different districts. Center-left parties reap greater electoral rewards when they offer descriptive representation, and that they also experience fewer difficulties in recruiting ethnic minority candidates. Therefore we argue that center-left parties have a greater incentive and ability to make their nominations more responsive to district demographics. More specifically, our hypothesis is that district-level ethnic diversity will increase the probability that any party will nominate an ethnic minority candidate, but this increase will be greatest for center-left parties. We look at multiple elections in Australia, the UK, and the US, and find consistent evidence in favor of this hypothesis. Even when center-left and center-right parties are nominating similar overall numbers of ethnic minority candidates, center-left parties’ descriptive representation patterns are more closely connected to district demographics. We argue that this helps explain how descriptive representation effects political competition more broadly.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Benjamin David Farrer is an assistant professor in the Environmental Studies Department at Knox College. His main research and teaching interests are in the field of parties and interest groups.

Joshua N. Zingher is an Assistant Professor of Political Science at Old Dominion University. His research focuses on mass political behavior, elections, and representation. His work has appeared in a number of journals including the British Journal of Politics Science, Electoral Studies, Party Politics, and the Journal of Politics.

ORCID

Benjamin David Farrer http://orcid.org/0000-0002-5889-5498

Joshua N. Zingher http://orcid.org/0000-0003-3928-4269

Notes

1 The Australian Labor Party, the British Labour Party, and the Democratic Party in the US all fall into this social-democratic party family (per the Comparative Manifestos Project coding), as does the Social Democratic Party of Germany and the New Democratic Party in Canada.

2 The Australian Conservative Party, the British Conservative Party, and the Republicans in the US all fall into the conservative or center-right party family, as does the Conservative Party in Canada and the New Zealand National Party.

3 We are assuming the existence of vote-seeking actors who decide nominations, who may be activists, party leaders, or party officials, depending on the nomination rules of the party (Norris and Lovenduski 1997; Kernell 2015).

4 Following disciplinary conventions, we use the terms “Black” and (in the US context) “African-American” interchangeably in this paper. We also use “Hispanic/Latino” to refer to individuals of Spanish-speaking heritage. Although “Hispanic/Latino” does not separate Spanish-speakers from individuals with Latin-American heritage, our key independent variable measuring this ethnicity comes from the US census, so we follow their coding and use the census terminology of “Hispanic/Latino”. Another potential label would be Latinx, to more explicitly incorporate individuals with a non-binary gender identity.

5 They may even have the opposite incentive. The districts with the greatest proportions of ethnic minority voters are often some of the safest seats for center-left parties. This implies that when center-right parties nominate candidates in heavily minority districts, these candidates are likely to lose (Sobolewska 2013). Thus, center-right parties likely have to nominate minority candidates in less diverse, more safely conservative districts if they wish to have minority candidates actually win. This is a plausible goal because it can improve national coverage (Black and Hicks 2006; Tavits and Cheng 2011), and can be part of an effort to “politically mainstream” race (Saggar and Geddes 2000; Sobolewska 2013).

6 As we discuss later, the issue of supply in specific districts is likely the most acute in the US case, due to the fact that candidates must live in the district where they run.

7 When nominations are centralized, parties could then take full advantage of the electoral payoffs from descriptive representation, by assigning candidates to districts where the ethnic demographics will be an asset. Recruitment difficulties in particular districts are likely to pose less of a problem under such systems.

8 Our case selection here limits the analysis to Anglophone industrial democracies with majoritarian electoral systems, following a most-similar cases principle. We also investigated Canada, but were only able to obtain data from one election. This makes the results non-comparable because we are less able to control for district-specific and election-specific fixed effects. In the interests of transparency, we nevertheless provide these results in the online appendix. Although we find that parties are responsive to district demographics, the lack of multi-year data makes it difficult for us to detect differences between the parties – although, if anything, the results run counter to our hypotheses. Further research could examine Canada in more detail to see if it is truly an outlier case.

9 In the empirical section, we also refine the analysis to focus only on one specific ethnic minority group at a time.

10 See: http://www.aph.gov.au/About_Parliament/Senate/Research_and_Education/platparl/c03.

11 Interestingly, Sobolewska (2013) notes that the Conservative Party was well of aware of its struggles with Britain’s ethnic and religious minorities during the first decade of the twenty-first century. However, even when the Conservatives attempted to reach out, they did not offer descriptive representation in heavily ethnic minority districts:

By 2010, the Conservatives realized that a more inclusive image was necessary to attract a broader base of supporters. Tory leadership actively sought to attract and nominate minority candidates in an effort to improve their image. However, since the majority of heavily minority constituencies are safe Labour seats the Conservatives are forced to nominate minority candidates in districts with small minority populations if they want to have a respectable chance of becoming an MP. (Sobolewska 2013, 623)

12 While the creation of majority–minority districts is ostensibly aimed at boosting minority representation in Congress, some have made the argument that it packs Democratic voters into a single district making the surrounding districts safer Republican seats (Hill 1995).

13 In the US case, we check our candidate ethnicity coding against that of Fraga (2016). With the exception of two cases, our coding is identical.

14 Another general trend is that even in the most representative parties, the percentage of minority candidates lags behind the percentage of ethnic minorities in the population.

15 Since the politically significant minorities are very different in each country, we eschew a potential “pooled” model of all countries.

16 We also present additional models in the appendix which include a lagged dependent variable. This helps us pick up autocorrelation due to both incumbency and relatively static district demographics. These models return substantively similar results that again support our hypothesis. An alternative approach would be to run separate regressions for each election year. We opted for pooling the data and then controlling for temporal effects, because separate regressions for each year would be akin to saying that what happened at election t is uninformative with respect to what happened at election t + 1, which we do not believe is a realistic assumption. Table 1, for example, demonstrates that there is a considerable degree of consistency over time. Therefore we pool the years together and use control variables to account for differences between election-years.

17 We do not have enough observations to investigate only open seats.

18 Unfortunately, we are unable to conduct this analysis on the British or Australian samples due to the fact that the respective censuses do not break the visible minority population down on the constituency level.

19 Another possible explanation is that the size of the legislature (150 seats) is much smaller in Australia than in the other three countries, making nominations an especially scarce resource.

20 A similar but less pronounced process might also be occurring in the exploratory analysis we conducted for Canada. There, the visible minority population is also quite diverse and the results were also imprecise, suggesting patterns in the opposite direction of our hypothesis. But again, this was with only one year of data for Canada, so results could be an artefact of election-specific factors, such as retiring incumbents leaving behind a unique crop of open seats.