The gendered nature of ethnic minority candidate nomination: An analysis of the moderating role of district diversity, party affiliation and ethnic minorities’ visibility

ABSTRACT

This paper investigates when party selectors prefer ethnic minority female over male candidates on their lists. We argue that the two competing strategies parties can follow (gaining the ethnic vote (‘ethnic community’ argument) versus not scaring ethnic majority voters (‘deterrence’ argument)) are dependent upon party ideology, district diversity and ethnic minorities’ visibility. Focusing on the 2018 local elections in Flanders (Belgium), we find that female Turkish/Maghrebi (i.e., visible ethnic minority) candidates outnumber their male counterparts in less diverse districts and rightist parties, whereas an opposite picture emerges in more diverse districts and leftist parties. Surprisingly, however, female candidates with less visible minority backgrounds outnumber their male counterparts in all contexts. These results imply a confirmation of the ‘ethnic community argument’, while casting doubts on the ‘deterrence’ argument. Taken together, our findings clearly highlight the conditionality of the selection of ethnic minority male versus female candidates in Proportional Representation list-systems.

KEYWORDS:

• candidate selection
• ethnic minorities
• intersectionality
• parties
• representation
• local elections

Introduction

Democratic institutions are increasingly expected by politicians, the public and scholars alike to mirror the diversity in society (Childs and Cowley 2011). Nevertheless, prominent groups such as women and ethnic minorities remain deprived of considerable political power. By diminishing the perceived legitimacy of political institutions (Ruedin 2013), reducing political trust and participation (Voicu and Comşa 2014), and impeding the translation of the interests of under-represented groups to the political stage (Phillips 1995), the descriptive under-representation of social groups challenges the functioning of democratic institutions and also sends a message of exclusion to both majority and minority groups (Phillips 1995). Conversely, including marginalised groups in democratic institutions may reduce political alienation, contribute to social stability, enhance the (perceived) quality of democracy, affect how societies perceive group members and influence how group members see themselves (Hughes 2013).

The behaviour of party selectors is crucial in this respect (Brouard and Tiberj 2010; Kittilson and Tate 2005; Shah 2014; Tolley 2019). While the presence of women (Bjarnegård and Kenny 2016; Vandeleene 2014) and of ethnic minorities (Akhtar and Peace 2019; Tolley 2019) on candidate lists has been investigated before, this is less the case for ethnic minority women. The intersectional perspective argues that intersectional groups have specific needs and interests that do not necessarily overlap with those of the separate groups (Hughes 2011). In this paper, we touch upon the debate regarding the ‘double jeopardy’ hypothesis stating that ethnic minority women in politics experience greater prejudice than ethnic minority men or ethnic majority women, as barriers associated with a minority group membership accumulate (Black 2000; Celis and Erzeel 2017; Janssen, Erzeel, and Celis 2021; Mügge and Erzeel 2016). Understanding these dynamics is crucial for designing inclusive and equitable recruitment processes that promote fair representation for all (intersectional) groups.

Parties’ strategic selection of male or female ethnic minority candidates depends on the trade-off between attracting ethnic minority voters and ethnic majority cosmopolitan voters who are supportive towards (certain) ethnic minority candidates, on the one hand, (hereafter the ‘ethnic community’ argument), and (not) deterring ethnocentric ethnic majority voters who are dismissive or even hostile towards (certain) ethnic minority candidates, on the other hand, (Dancygier 2017; English 2020; Jacobs, Camille, and Jean-Benoit 2013) (hereafter the ‘deterrence’ argument). Following previous research (see, for example, Dancygier 2017; Janssen, Erzeel, and Celis 2021), we argue that the choice between these two strategies (and thus the choice to select primarily male or female ethnic minority candidates) depends on the district diversity and parties’ ideological stance.¹ While almost all studies on intersectional inclusion up to now either focused on the national level (Celis and Erzeel 2017; Celis et al. 2014; Mügge and Erzeel 2016) or on local elections in an urban context (Azabar and Thijssen 2021; Celis, Eelbode, and Wauters 2013; Dancygier 2017; Dowley 2021; Janssen, Erzeel, and Celis 2021), we examine all 309 local districts in Flanders (300 municipalities and 9 city districts within Antwerp, the biggest city in Flanders), ranging from very rural and ethnic homogenous districts to urban and highly diverse districts, enabling to fully test the effect of district composition. This paper, furthermore, takes into account the variation between visible (i.e., Turkish/Maghrebi candidates) and less visible minorities (i.e., candidates of other non-Western origin). We assume that both the ‘ethnic community argument’ and the ‘deterrence argument’ will have more impact for visible minorities than for less visible ones as visible ethnic minorities are more readily identifiable as members of minority groups and may face specific challenges related to discrimination, stereotyping and exclusion.

In sum, our central research question reads as follows ‘which factors moderate the gender gap in the selection of ethnic minority candidates in Flanders?’. We hereby focus on three potential moderating factors, i.e., 1) the ethnic composition of the district, 2) the ideological positioning of political parties nominating candidates and 3) the visibility of the ethnic minority group.

If the ‘ethnic community’ argument (trying to attract ethnic minority voters) prevails, party selectors will opt for male candidates, who are generally better vote mobilisers among ethnic minority voters (Akhtar and Peace 2019; Dancygier 2017; Janssen, Erzeel, and Celis 2021). We expect this effect to be more outspoken in leftist parties, in ethnically diverse districts and for visible minorities. When trying not to deter ethnic majority voters (the ‘deterrence’ argument), female candidates constitute a safer choice towards ethnic majority voters (Celis and Erzeel 2017; Garrett 2017). We expect the preference for ethnic minority women over ethnic minority men to be more outspoken in rightist parties, in less diverse districts and for visible ethnic minorities.

Our results reveal that as expected, party selectors select more visible ethnic minority women than men in less diverse electoral districts and in rightist parties, whereas the opposite is true in highly diverse districts and leftist parties. However, less visible female candidates are preferred over less visible male candidates in all kinds of parties and districts, which goes against our expectations. This puts doubts on the plausibility of the ‘deterrence’ argument.

The gender gap among visible ethnic minority candidates

For the diversification of politics to be successful, the behaviour of political parties and their selectors is crucial (Brouard and Tiberj 2010; Kittilson and Tate 2005; Shah 2014; Tolley 2019).² On the one hand, parties are considered as racialised and gendered organisations who limit opportunities for ethnic minorities and women (Kenny 2013; Murray 2015). On the other hand, political parties, incentivised by structural measures and societal changes, increasingly take action to address the underrepresentation of these groups (Bird, Saalfeld, and Wüst 2011; Hughes 2011).

Although ethnic parties are often considered to be the main vehicle driving the representation of ethnic minorities (Ishiyama and Stewart 2021), there are many instances in which ethnic parties are either non-existent or insufficiently developed to be strong competitors in elections. Under these circumstances, the representation of ethnic groups can be ensured by majority political parties, who could possibly expand their electorate with ethnic voters (Buta and Gherghina 2022).

When drawing up their candidate lists, party selectors make a rational calculus and try to select candidates that are both able to attract voters and that are least likely to cost voters to the party. It is commonly thought that group members whose perceived electability is higher and whose profile is less threatening to the power status-quo have higher chances of being selected for joining the political class (Durose et al. 2013). In this respect, studies increasingly take into consideration the intersection of ethnicity and gender in candidate nomination procedures (Janssen, Erzeel, and Celis 2021). Previous research has in this regard shown that party selectors have a baseline preference for visible ethnic minority women over ethnic minority men (Celis and Erzeel 2017; Celis et al. 2014). Not only are men traditionally perceived as more concerned with power strategies than women (Schneider et al. 2016), which implies that ethnic minority men constitute a greater challenge to ethnic majority (male) incumbents, but ethnic minority female candidates also have a complementary advantage. They enable party selectors to simultaneously diversify their list in terms of gender and ethnicity, which is especially interesting when parties need to comply with gender quota (Celis and Erzeel 2017; Dancygier 2017; Garrett 2017).

Furthermore, visible ethnic minority women have an extra advantage over ethnic minority male candidates because they are more acceptable to the ethnic majority electorate. Party selectors often fear that voters associate visible ethnic minority male candidates with criminality, terrorism and other perceived negative side-effects of immigration they want to avoid (Celis and Erzeel 2017).

Visible ethnic minority female candidates, on the contrary, are more likely to be perceived as examples of well-integrated citizens because simply by taking up a public role they send out the message that they renounce the patriarchal societies in which they or their ancestors grew up. These impressions become even more outspoken when visible ethnic minority female candidates refrain from wearing religious or other symbols such as a headscarf (Dancygier 2017; Garrett 2017). Such candidates are thus likely to attract cosmopolitan voters and least likely to defect ethnocentric majority voters.

In this study, party selectors’ supposed baseline preference for female visible ethnic minorities is examined by looking at the presence of Turkish/Maghrebi candidates on Flemish local candidate lists. In Belgium, the Turkish and Maghrebi communities constitute the most important visible ethnic minority groups (Dowley 2021; Janssen 2020), whose gendered representation has shown to be the subject of strong inclusion dilemmas in urban cities across the country (Dancygier 2017; Janssen, Erzeel, and Celis 2021).

Following previous research, we argue that the general advantage of visible ethnic minority women is not set in stone, but dependent on party ideology and district diversity (Janssen, Erzeel, and Celis 2021; Mügge 2016). Parties’ ideological position and the diversity of the electoral district determines what kind of electorate parties are likely to attract, i.e., how parties’ specific electorates are composed in terms of the proportion of cosmopolitan ethnic majority voters, ethnocentric ethnic majority voters and ethnic minority voters. Depending on the respective number of ethnic minority voters, cosmopolitan ethnic majority voters and ethnocentric ethnic majority voters the party aims to serve, selectors will be likely to prefer either ethnic minority women or men on their lists.

In urbanised and highly diverse districts, ethnic minority male candidates appear to have an advantage over their female counterparts (Janssen, Erzeel, and Celis 2021). In these contexts, ethnic minority voters are highly numerous and generally outnumber potential ethnocentric voters who are most critical against ethnic minority candidates. Therefore, parties seeking electoral gains will thus select ethnic minority candidates who are (expected to be) most preferred by ethnic minority voters. Ethnic minority voters, and more specifically the Muslim electorate, are known to support male candidates at higher rates than female candidates. Immigrants from different countries are often thought to have divergent gender role attitudes shaped by their distinct home countries’ cultural traditions and subsequent migration experiences (Maliepaard and Alba 2016; Wang and Coulter 2019). Previous research in this regard highlights that Muslim immigrants are more likely to hold conservative gender views (Alexander and Welzel 2011; Janssen, Erzeel, and Celis 2021; la Roi and Mood 2022; Ross 2008) and to support patriarchal values, even if they are second- or third-generation immigrants (Alexander and Welzel 2011; Maliepaard and Alba 2016; Spierings 2015). As a result, men dominate the public life in ethnic minority communities and, therefore, have the biggest networks and enjoy the highest social standing. As such, ethnic minority male candidates are expected to be better vote mobilisers and are, therefore, the preferred type of candidate in these districts (Akhtar and Peace 2019; Dancygier 2017; Janssen, Erzeel, and Celis 2021). Empirical evidence for ethnic minority male candidates’ electoral advantage are, however, mixed (Azabar and Thijssen 2021; Dancygier 2017; Philpot and Walton 2007; van der Zwan, Tolsma, and Roos 2020).

H1:

Ethnic diversity at the district level favours the presence of male ethnic minority candidates over female ethnic minority candidates on the list

When it comes to the moderating effect of party ideology, there is vast literature showing that leftist parties are more likely to nominate women (Caul 1999; O’Brien 2018) and higher numbers of ethnic minority candidates (Eelbode et al 2013; van der Zwan et al. 2019). However, our expectations differ significantly for the intersectional group of ethnic minority women. Unlike the general trend, the advantage for ethnic minority women is expected to be more pronounced among rightist parties than among leftist parties.

To start, rightist parties are likely to attract a fair share of ethnocentric voters, whereas both ethnic minority voters and cosmopolitan voters are skewed to the left (Jacobs et al. 2016). Therefore, when rightist parties select ethnic minority candidates, they will be likely to choose those candidates who are the least likely to offset ethnocentric voters (English 2020). As outlined above, selecting ethnic minority female candidates is in these circumstances the safest choice. Leftist parties, in turn need to find a balance between the preferences of ethnic minority voters and cosmopolitan ethnic majority voters. The former are likely to prefer ethnic minority male candidates, as explained above. The latter, however, adopt more liberal views on, for instance, gender, sexuality and diversity, and are likely to have difficulties with ‘illiberal’ ethnic minority (i.e., male) candidates. Again, ethnic minority female candidates are expected to be a safer choice in this regard. Given this trade-off between attracting ethnic minority voters and cosmopolitan voters, leftist parties are likely to compose more gender-balanced lists than their rightist counterparts.

H2:

Rightist parties favour the presence of female ethnic minority candidates over male ethnic minority candidates on the list.

The gender gap among less visible ethnic minority candidates

The distinction between visible and less visible ethnic minorities is crucial for a comprehensive analysis of the gender gap among ethnic minority candidates. How ethnic minorities’ visibility should be defined is subject of fierce discussion (Song 2020). We define visible minorities based on the outlook (phenotype) of the members of the ethnic group, the group’s size and the distinctiveness of their cultural norms and practices (Garrett 2017). Visible ethnic minorities often belong to groups that exhibit distinct physical or cultural markers, are more readily identifiable as members of minority groups and may face specific challenges related to discrimination, stereotyping and exclusion (Bloemraad 2013; Kariv, Menzies, and Brenner 2010). Consequently, it is the representation of these visible minorities which is the most politicised, as they are both subject to political and societal debates, and a source of votes to be won in elections (Janssen 2021). Considering that Turkish and Moroccan minorities are the most numerous ethnic groups, that they are quite easily recognisable, and that they are more likely to hold more conservative (gender) views (Alexander and Welzel 2011), Turkish and Moroccan candidates were selected as visible ethnic minority candidates.

Less visible minorities, on the contrary, are less numerous and weigh, hence, less heavily on the total number of votes in a district. The incentive to nominate male candidates to attract the ‘ethnic vote’ (as explained above) is therefore less strong, rendering the ‘ethnic community’ argument less pertinent.

Furthermore, ethnic minority voters originating from more gender equal societies are less likely to grant a dominant role in social relations to men than ethnic minority voters stemming from patriarchal communities (Alexander and Welzel 2011; Janssen, Erzeel, and Celis 2021; la Roi and Mood 2022; Ross 2008). This should reduce the supposed advantage of male candidates in highly diverse districts. Additionally, the trade-off leftist parties face between attracting ethnic minority voters by nominating male ethnic minority candidates and not deterring ethnic majority voters by nominating female ethnic minority candidates is expected to be lower when less visible ethnic minority candidates are concerned. Because leftist parties and their voters generally have (more) positive attitudes towards ethnic minority candidates (than their rightist counterparts) (Besco 2020; Ceyhan 2018; Van Trappen, Devroe, and Wauters 2020), they can be expected to be steered towards ethnic minority female candidates only or primarily when ethnic minority male candidates appear to be illiberal and thereby contradict their progressive beliefs (Dancygier 2017).

Finally, ethnocentric voters can equally be expected to be less opposed to male candidates stemming from less visible and stigmatised groups because of their lower associations with criminality and terrorism. Therefore, rightist parties are equally more likely to compose gender balanced lists in terms of the selection of candidates of less visible ethnic minority backgrounds.

In sum, both the ‘ethnic community’ argument and the ‘deterrence’ argument are less applicable to less visible minorities. We, therefore, hypothesise that candidate lists will be gender balanced for them, irrespective of the level of district diversity or the party ideology:

H3:

Candidate lists are gender balanced in terms of candidates of less visible ethnic minority backgrounds irrespective of the level of district diversity or the ideological affiliation of the political party

Case selection

This paper studies the intersectional inclusion of ethnic minority male and female candidates on candidate lists in Flanders (the Dutch-speaking part of Belgium). Almost all studies on intersectional inclusion up to now have either focused on the national level (Celis and Erzeel 2017; Celis et al. 2014; Mügge and Erzeel 2016) or on local elections in an urban context (Azabar and Thijssen 2021; Celis, Eelbode, and Wauters 2013; Dancygier 2017; Dowley 2021; Janssen, Erzeel, and Celis 2021). We extend the scope by examining all 309 local districts in Flanders (300 municipalities and 9 city districts within Antwerp, the largest city in Flanders), ranging from very rural areas with almost no ethnic minority inhabitants to urban areas with a highly concentrated ethnic population, which enables us to fully test the effect of district composition.

The Flemish local context is, furthermore, an interesting case because its socio-demographic, political and institutional features provide strong opportunities for the promotion of gender equality and ethnic diversity. The local level is considered to be the most accessible level for marginalised groups (Bloemraad 2013) and ethnic organisations and networks have more impact on selection processes at the local level compared to higher levels (Akhtar and Peace 2019). Additionally, the Flemish proportional representation system with semi-open lists is also considered to foster political diversity (Norris 2004) as it grants parties the opportunity to balance their ticket. This allows party selectors to select new, diverse candidates whilst preserving seats for (ethnic majority male) incumbents. The strong preferential voting system also incentivises candidates to pursue a personal vote and parties to nominate candidates with ties to ethnic minority communities to maximise their electoral score. That non-Belgian EU-citizens enjoy voting rights for local elections since 2000 and non-EU citizens since 2004 further provides local parties with extra incentives to include immigrant-origin candidates on their lists. Flanders can also be characterised as a success story in terms of women representation partly due to the stringent gender quota system (in place since the 1990s), also at the local level.

The candidate selection process at the local level is dominated by the local party chair and the local executive committee. In most cases, the local party elite drafts a candidate list, including a predefined list order in which the candidates will appear. The model list is presented to the local party members who can either approve or disapprove this model list (De Winter et al. 2013).

Data collection

The 2018 local elections in flanders

This study relies on an examination of the candidate lists participating in the 2018 local elections in Flanders. First, we delineated the research population to all lists which could be affiliated with a single national party, whether it be based on the name of the party list or on information found in online newspapers or on party websites. Independent local lists were not coded because of feasibility reasons. Joint lists were coded when at least one of the participating parties was affiliated with a national party. However, because the ideological positioning of joint lists is difficult to determine, joint lists were only included in the robustness check, as is discussed below. Second, as no government statistics are available about candidates’ ethnic background, their ethnic origin was coded manually following a conventional approach (Janssen 2020). As Dutch and French are the main languages used in Belgium,³ we explored the migration background of candidates with a non-Dutch or non-French surname and/or forename. This was done by browsing through online news websites, official party and candidate websites and social media accounts. When the information found online clearly confirmed that the candidates had a foreign origin, they were assigned to either one of two categories. The first category only includes candidates with a Turkish/Maghrebi origin, matching the approach of previous studies (Dancygier 2017; Dowley 2021; Janssen 2020; Janssen, Erzeel, and Celis 2021). The second category, ‘less visible ethnic minority candidates’, comprises candidates of non-European, Southern-European or Eastern-European origin, with the exception of candidates of Turkish/Maghrebi descent. It should thus be noted that candidates of Western-EU or Northern-EU origin were not coded as ethnic minority candidates because their political role is far less politicised in Belgium than candidates of other foreign origins (Celis and Erzeel 2017).

Subsequently, the candidates for whom no concrete information was found online were coded based on the geographical presence of their name, as depicted on the genealogy portal ‘Forebears’ (http://forebears.co.uk). When their surname and/or forename was predominantly present in the Turkey/Maghreb countries, they were coded as Turkish/Maghrebi candidates. When their surname and/or forename was predominantly present in other foreign regions, they were coded as less visible ethnic minority candidates. For all ethnic minority candidates, their gender was coded as well.

The entire coding process was supervised by one of the authors, who also coded the vast majority of the candidate lists. Two other researchers and two student assistants each coded a smaller part of the candidate lists. Intercoder reliability was guaranteed as random samples of the original (uncoded) candidate lists assigned to each coder were re-coded by one of the others. Both coded lists were subsequently compared and in case differences occurred, it was discussed which code should be given. The candidates whose ethnic origin remained uncertain were coded independently by all authors, after which they collectively came to a deliberate decision.

In the end, 267 lists were identified as comprising at least one candidate of Turkish/Maghrebi descent and were used to test H1–2. 505 lists included at least one less visible ethnic minority candidate which enabled us to test H3.

Variables

Two dependent variables are included in this study: the ratio of female to male candidates of Turkish/Maghrebi descent (H1 & H2) and the ratio of female to male candidates with less visible ethnic minority backgrounds (H3).⁴ The values of these continuous variables rank from 0 (lists including only ethnic minority male candidates) to 1 (lists including only ethnic minority female candidates). A value of 0.5 means that an equal number of ethnic minority male and female candidates were selected. Values between 0 and 0.5 indicate that the participants selected more male than female candidates, whereas values between 0.5 and 1 imply the opposite (for a similar approach, see Janssen et al. (2021)).

The first independent variable, ‘party affiliation’, distinguishes between leftist, centrist and rightist parties. Candidate lists associated with the far left party (PVDA), green party (Groen) and social-democratic party (sp.a) were categorised as leftist. Party lists of the Christian-democratic party (CD&V) were coded as centrist. Rightist candidate lists included lists affiliated with the regionalist party (N-VA), the liberal party (Open Vld) and the far right party (Vlaams Belang). Two other independent variables measured the diversity of the electoral district by looking at the size of the Turkish/Maghrebi population and the size of the less visible ethnic minority population⁵ (Vlaanderen 2021; Stad Antwerpen 2020).

As control variables, we include openness towards ethnic minorities in the electoral district by looking at the vote share of the far right party, Vlaams Belang, in the previous (2012) elections. The electoral strength of the far right party could steer other parties to selecting Turkish/Maghrebi female rather than male candidates, because female candidates are expected to be a safer choice (Celis and Erzeel 2017). All data were derived from the official government website (Agentschap Binnenlands Bestuur, 2018). Additionally, we controlled for the proportion of Turkish/Maghrebi or less vibible ethnic minority candidates on the list. Finally, the size of the local council was taken into account. The more candidates parties can select, the more likely they are to balance their ticket (Norris 1997).

Statistical tests

To estimate the gender ratio of visible and less visible ethnic minority candidates, we conducted fractional regression analysis. The main text discusses two models based on the candidate lists affiliated with one national party (cartel lists excluded) and measure parties’ ideological profile as a categoric variable as discussed above. As robustness check, we conducted additional analyses which included the cartel lists, thereby measuring parties’ ideological profile by their exact party affiliation with cartel lists constituting a separate category.

Results

The gendered selection of Turkish/Maghrebi candidates

For the first part of our analysis, we focus on the gendered representation of Turkish/Maghrebi candidates on the candidate lists of the 2018 local elections. A descriptive overview of these candidate lists shows that there was no profound gender gap among Turkish/Maghrebi candidates. 45% of the candidate lists including at least one candidate of Turkish/Maghrebi descent counted more male than female Turkish/Maghrebi candidates, whilst 46% of these lists comprised more female than male Turkish/Maghrebi female candidates. 9% included an equal number. The average ratio of female Turkish/Maghrebi candidates on a list was 0.52. A one-sample t-test showed that this is not statistically different from 0.5 (t(266) = 0.95, p > 0.1).

In Table 1, a fractional regression model tests whether the ratio of female Turkish/Maghrebi candidates is dependent upon party ideology and district diversity. To start, rightist parties have a significant higher ratio of female Turkish/Maghrebi candidates than their leftist counterparts, which confirms H2. Furthermore, the size of the Turkish/Maghrebi population has a statistically significant negative effect. This effect is plotted in Figure 1, which clearly confirms H1. Female Turkish/Maghrebi candidates are preferred over their male counterparts in the least diverse districts, whereas male Turkish/Maghrebi candidates are preferred in the most diverse districts. None of the control variables has a significant effect.

Figure 1. Predicted values of the ratio of female Turkish/Maghrebi candidates, by size of the Turkish/Maghrebi population.

Table 1. Fractional regression model estimating the ratio of female Turkish/Maghrebi female candidates.

	b (SE)
Intercept	0.11 (0.46)
Party type (ref. leftist)
Centrist	−0.04 (0.28)
Rightist	0.47 (0.27)^^
Size Turkish/Maghrebi population	−0.05 (0.02)**
Proportion Turkish/Maghrebi candidates on list	0.02 (0.02)
Vote share far right in 2012 elections	−0.00 (0.03)
Size municipal council	0.01 (0.01)
D squared (adjust)	0.02
N	267

Significant codes: p < 0.001=***, p < 0.01=**, p < 0.05=*, p < 0.1=^{^}.

The robustness check in Appendix 1.1 confirms that the size of the Turkish/Maghrebi population exerts a negative significant effect, but shows more nuanced effects of party affiliation. To start, compared to the far left party, PVDA, all other parties had a lower ratio of female Turkish/Maghrebi candidates, but only the differences between far left on the one hand and social-democratic candidate lists and cartel lists on the other hand were statistically significant. This contradicts to a certain extent H2, as it was a leftist party (PVDA) which had the highest ratio of female Turkish/Maghrebi candidates. The difference between leftist and rightist parties found in Table 1 is thus primarily driven by social-democratic parties, which had the lowest ratio of female Turkish/Maghrebi candidates.

The gendered selection of candidates with less visible ethnic minority backgrounds

In contrast to the selection of Turkish/Maghrebi candidates, there is a clear gender gap in the selection of candidates with less visible ethnic minority backgrounds. 56% of the candidate lists comprised more less visible ethnic minority female than male candidates, whereas only 30% comprised more male than female candidates. 14% of the lists contained an equal number. The average ratio of less visible ethnic minority female candidates amounted to 0.62. A one-sample t-test showed that this value is statistically different from 0.5 (t(504) = 6.67, p < 0.01). Thus, contrary to our expectations (H3), the selection of less visible ethnic minority candidates is more gendered than the selection of highly visible ethnic minority candidates.

We examined the origin of these less visible ethnic minority candidates more closely. A distinction was made between candidates stemming from Southern-Europe, Eastern-Europe, South-Eastern-Europe, Central- and Latin-American origin, other Asian countries and other African countries (i.e., excluding Turkey and Maghreb countries). In all categories, female candidates outnumbered their male counterparts (Appendix 2). The differences were especially large for candidates stemming from (South-)Eastern-Europe and Central- and Latin-America.

The fractional regression analysis (Table 2), furthermore, shows that the advantage of less visible ethnic minority female candidates emerges in leftist, centrist and rightist parties as well as in parties running in less and more diverse districts. Figure 2 shows that even in the most diverse municipalities, female less visible ethnic minority candidates outnumber their male counterparts. Taken together, hypothesis 3 is, therefore, clearly rejected.

Figure 2. Predicted values of the ratio of female less visible ethnic minority candidates, by size of the less visible ethnic minority population.

Table 2. Fractional regression models estimating the ratio of female candidates with less visible ethnic minority backgrounds.

	b (SE)
Intercept	0.78 (0.33)*
Party type (ref. leftist)
Centrist	0.26 (0.21)
Rightist	0.12 (0.17)
Size less visible ethnic minority population	0.00 (0.01)
Proportion less visible ethnic minority candidates on list	−0.00 (0.01)
Vote share far right in 2012 elections	−0.02 (0.02)
Size municipal council	−0.01 (0.01)
D squared (adjust)	0.00
N	505

Significant codes: p < 0.001=***, p < 0.01=**, p < 0.05=*, p < 0.1=^{^}.

The robustness check (Appendix 1.2) equally shows that there is no significant influence of the size of the less visible ethnic minority population. Regarding party differences, both Christian-democratic party lists, far right party lists and cartel lists had a higher ratio of female candidates with less visible ethnic minority backgrounds than far left parties.

Again, we further examined these trends per party more closely by looking at the region of origin of the less visible ethnic minority candidates (Appendix 3). The results show that all parties selected more female than male candidates for almost all ethnic minority categories. When looking at differences between the least diverse districts (<14%), moderate diverse ([14–29%]) and the most diverse districts (≥29%) (Appendix 4), it shows that female candidates outnumber male candidates in all but one category. In the most diverse districts, male candidates of Southern-European origin outnumber their female counterparts.

Discussion and conclusion

This paper contributed to the scholarly understanding of the intersectional (dis)advantages faced by ethnic minority women and men aspiring candidacy for elections. We tackled two shortcomings of previous research on this topic. To start, by examining the gendered ethnic composition of the candidate lists participating in the 2018 local elections in all municipalities in Flanders (Belgium) this paper examined for the first time how gendered ethnic minority inclusion varies across a wide variety of local contexts, ranging from rural and ethnic homogenous districts to urban and highly diverse districts. Additionally, by paying attention to not only the most visible minorities groups (i.e., Turkish/Maghrebi candidates) but also less visible minority groups (i.e., candidates of other non-Western origin), this paper provided a nuanced image on the intersectional (dis)advantages faced by ethnic minority men and women.

For this analysis, we use two strategies that are likely to steer the gendered selection of ethnic minority candidates: attracting the ethnic vote (‘ethnic community’ argument) versus not deterring ethnic majority voters (‘deterrence argument’). Our results, however, only partly confirmed the relevance of these strategies. We demonstrated that the ‘ethnic community’ argument drives party selectors to select more male than female visible minority candidates in highly diverse districts. Furthermore, with the exception of far left parties, leftist (i.e., social-democratic and green) parties prefer visible ethnic minority men whereas rightist (i.e., liberal and regionalist) parties prefer visible ethnic minority women. This is remarkable given that women in general are better represented in leftist parties (O’Brien 2018). For less visible minorities, on the other hand, we showed that female candidates consistently outnumbered their male counterparts across all parties and all kinds of districts.

While our study sheds light on the gender disparities among ethnic minority candidates, it is important to acknowledge that the underlying factors driving these patterns require further exploration. Future research should, for example, delve into the origins of why female less visible ethnic minorities consistently outnumber their male counterparts. One potential explanation is that party selectors perceive (male) ethnic minority candidates as threatening to ethnic majority voters, regardless of whether they belong to more or less visible ethnic minority groups. However, we find this explanation less plausible for two reasons. To start, traditional news media in Flanders reporting on crime news typically portray Maghrebi citizens as more threatening that citizens of other ethnic minority backgrounds (Jacobs 2017). It is, therefore, unlikely that party selectors will perceive male less visible ethnic minority candidates as equally threatening than male visible ethnic minority candidates. Furthermore, this argument is insufficient to explain why in ethnically diverse districts, male visible ethnic minority candidates outnumbered female visible minority candidates, but an opposite picture emerged among less visible ethnic minority candidates. These differences demonstrate that there are different mechanisms steering the gendered selection of more and less visible minorities.

Second, as both visible and less visible female ethnic minority women outnumber ethnic minority men on the candidate lists, the theoretical expectations based on demand factors potentially do not suffice and should be complemented by theoretical assumptions about supply-side drivers. For instance, future research could explore the possibility that women belonging to less visible ethnic minority communities are more likely to come forward as aspirants compared to women from visible minorities. Whereas previous research (Akhtar and Peace 2019) has shown that in conservative ethnic minority communities female aspirants are often discouraged to run, the opposite might be the case among ethnic minority women stemming from less patriarchal and conservative societies.

Furthermore, scholars should pay attention to how gender quota legislation, e.g., demanding that candidate lists must include as many women as men as in Flemish local elections, might amplify the impact of these supply factors. The requirement for candidate lists to include a particular share of women may incentivise party selectors to select female aspirants belonging to less visible minorities, as they enable parties to kill two birds with one stone, i.e., simultaneously meeting gender quota requirements and enhancing ethnic diversity.

It is, furthermore, important to acknowledge that our study is subject to certain limitations and potential sources of omitted variable bias. Specifically, we recognise that additional factors such as the number of competitors, political experience of candidates and the size of the voting population (both ethnic majority and ethnic minority voters) might have an impact on our findings. Therefore, we encourage future research endeavours to explore these factors in greater depth as their inclusion would likely provide valuable insights into the dynamics at play.

In sum, our results seem to confirm the ‘ethnic community’ argument for visible ethnic minorities (if an ethnic minority is numerous in a district, ethnic minority men have a better chance to be selected), but put doubts on the ‘deterrence’ argument (it is not the case that women only have an advantage among highly visible ethnic minorities). Therefore, this study calls for future work to address parties’ demand together with ethnic minority aspirants’ motivations but also voters’ preferences, as the supply and demand for minority candidates are closely intertwined in shaping the political inclusion of minority groups in politics (Janssen 2020). This will enable scholars to come to a better understanding of the complementarity advantage attributed to female (visible) ethnic minority candidates.

While our study focused specifically on the context of Flanders, the implications of our findings hold relevance for broader discussions on gender and ethnic representation. Future research endeavours should explore the generalisability of our findings in diverse settings, especially those without electoral quota provisions, to further disentangle the mechanisms steering the strategic selection of ethnic minority male and female candidates. As the position of both women and ethnic minority groups in society and in politics in Flanders is comparable to that in many other Central and West-European countries and as the Flemish electoral system share some common features with those of other countries, we expect our findings to travel well across similar systems in Europe. It is, in this respect, worth noting that our findings align with observations in neighbouring countries, where ethnic minority women have demonstrated a greater propensity to ascent the political ladder compared to their male counterparts. A study conducted in France, for example, highlights that minority women have witnessed a swifter advancement within the political elite than minority men (Garrett 2017). Equally, Freidenvall (2016) showed that the importance attached to the political inclusion of women in Sweden, counterbalanced a weaker trend towards the inclusion of minority groups, so that minority women are equally, or better, represented than minority men. This also comes to the fore in Hughes (2016) comparative study highlighting that Muslim ethnic minority women outnumber men in the Netherlands, Sweden and Norway. These parallel trends further underscore the importance of examining the specific dynamics at play. Understanding these dynamics is crucial for designing inclusive and equitable electoral practices that promote fair representation and participation for all individuals.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/03003930.2023.2239152.

Notes

1. Studies focusing on the political (under)representation of social groups differentiate between macro-, meso- and micro-level factors. Although both macro-level factors (variations between countries/electoral systems) and micro-level factors (related to individual voters and candidates) are relevant to consider, our article uses a meso-level approach taking into consideration variation at the party and district level.

2. Apart from parties’ selection strategies, a range of other factors, including the institutional context, the mobilising capacity of minority groups, and the role of voters, account for the political underrepresentation of diverse groups.

3. There is also a very small German-speaking part.

4. Additional analyses focusing on the relative list position can be found in Appendix 5. Following van der Zwan, Lubbers, and Eisinga (2019), the candidates’ relative list position was calculated as follows: $\frac{{\mathrm{t}}_{\mathrm{i}}-{\mathrm{p}}_{\mathrm{j}}}{{\mathrm{t}}_{\mathrm{i}}-1}$ for which ${\mathrm{t}}_{\mathrm{i}}$ stands for the total amount of candidates on list $\mathrm{i}$ and ${\mathrm{p}}_{\mathrm{j}}$ stands for the position of candidate $\mathrm{j}$ on list $\mathrm{i}$. Candidates’ relative list positions range from zero (last candidate) to one (first candidate). Our results point out that there is no statistically significant difference in the relative list positions assigned to female and male Turkish/Maghrebi, and female and male less visible ethnic minority candidates. These findings hold across both leftist, centrist and rightist parties and across different degrees of district diversity.

5. Citizens of both non-EU, Southern-European and Eastern-European origin were taken into account when calculating the size of the ethnic minority population, thereby excluding the population of Turkish/Maghrebi descent.