Representing their own? Ethnic minority women in the Dutch Parliament

Abstract Ethnic minority women tend to be better represented in parliaments than ethnic minority men. What does this mean for their substantive representation? This article makes use of intersectional analysis to study how the relationship between descriptive and substantive representation differs within and between gender and ethnic groups. Drawing on written parliamentary questions and the committee memberships of MPs in seven parliamentary sessions (1995–2012) in the Netherlands, a strong link is found between descriptive and substantive representation. Female ethnic minority MPs more often sit on committees and table questions that address ethnic minority women’s interests than male ethnic minority and female ethnic majority MPs. The link, however, is fragile as it is based on a small number of active MPs. This demonstrates the importance of an intersectional approach to understanding how representation works in increasingly diverse parliaments, which cannot be captured by focusing on gender or ethnicity alone.

'own' ethnic group on the political agenda? To paraphrase Mansbridge (1999): do ethnic minority women represent ethnic minority women?
Drawing on data from seven parliamentary sessions (1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012) in the Netherlands, we analyse which MPs addressed issues in parliamentary questions concerning the position of ethnic minority women. Among European immigration countries, the Netherlands has one of the strongest and longest traditions of electing ethnic minorities to political office (Fernandes et al. 2016), allowing us to study variation over time. Based on existing research on the political representation of women and ethnic minorities, the following section formulates two hypotheses. We then present a brief overview of the presence of ethnic minorities, their gender and party affiliation in the Dutch parliament, followed by our results. Although we find a strong empirical link between descriptive and substantive representation, the link is fragile as only a small number of MPs table questions that concern the position of ethnic minority women in society.

Descriptive and substantive representation
A key question in the study of the political representation of structurally under-represented groups is the relationship between descriptive and substantive representation. Descriptive representation refers to the numerical presence of particular groups in elected office; substantive representation to 'acting in the interests of the represented in a manner responsive to them' (Pitkin 1967: 209). The literature addresses three debates that speak to our research question.
A first strand of scholarship argues that descriptive representation leads to substantive representation: the greater numerical presence of women or ethnic minorities in parliament furthers their interests in policy-making (Mansbridge 1999;Williams 2000;Young 1990). Philips (1995: 62) argues that the presence of women will enhance the visibility of 'particular interests of women that would otherwise be overlooked'an argument that has found empirical support in numerous countries (e.g. Bird 2005;Bratton 2005;Chattopadhyay and Duflo 2004;Gerrity et al. 2007). Similarly, research on ethnic minority MPs finds them more inclined to champion the interests of ethnic minorities than their ethnic majority colleagues (Donovan 2012;Saalfeld 2011;Saalfeld and Bischof 2013). A key explanatory model here is that of 'critical mass': female MPs are unlikely to sponsor bills on traditional women's issues until there is a certain number or percentage of women in the legislature. 1 Until this tipping point is reached, women legislators are more likely to serve as tokens (Dahlerup 2006).
A second body of literature argues that descriptive representation does not necessarily lead to substantive representation (Weldon 2002). Authors here do not support the idea of a critical mass, but emphasise the importance of 'critical acts' (Dahlerup 1988); we should not simply focus on the presence of women but ask how substantive representation occurs and what specific actors do (Childs and Krook 2006). 'Critical actors'both female and maleseek to 'successfully or unsuccessfully, represent women substantively, as a group' (Childs and Krook 2009: 144). The mere presence of women or ethnic minorities in politics does not make a 'drastic' difference due to institutional and contextual barriers (Tremblay 1998). Female politicians rarely present themselves as 'women politicians' representing only women. For electoral reasons, they tend to present themselves as politicians wholike their male counterpartsrepresent the broader electorate (see Leyenaar 2004). Moreover, studies point to the mechanism of 'tokenism' where female representatives of colour are admitted to the corridors of power but are expected to embrace entrenched agendas (for the US, see Hawkesworth 2003;Hutardo 1996; for France, see Murray 2016). Finally, ethnic minority MPs must navigate conflicting expectations: their party expects them to represent voters regardless of their background, while members of their 'own' ethnic group often expect MPs to represent them especially (Celis and Wauters 2010;M€ ugge and Schotel 2017). Ethnic minority politicians thus do not 'automatically support' issues of concern for constituencies with whom they share similar backgrounds (Aydemir and Vliegenthart 2016).
A third, related debate questions the link between descriptive and substantive representation as it 'assumes a homogeneity among women that reinforces essentialist notions of an exogenously given, universally shared, fixed female identity' (Franceschet and Piscopo 2008: 396; also see Celis and M€ ugge 2018). Without denying that shared perspectives 'increase the likelihood that some shared interests can be articulated by group members', scholars here see the articulation of women's interests as a fluid process (Franceschet and Piscopo 2008: 397). Dovi (2012) argues that 'descriptive representatives' have no innate characteristics; what matters are strong mutual relationships with a particular group. Franceschet and Piscopo (2008: 397) distinguish between 'process-oriented' and 'outcomeoriented' aspects of representation. Substantive representation as a process 'occurs when legislators undertake activity on behalf of some or many women', for example introducing bills that address women's issues or putting women's issues on the agenda of committees with the aim of transforming policy. In this paper, we focus on the process-oriented aspects.
In line with the critical actors approach, the politics and gender literature has shown that not only women, but also men (claim to) promote women's interests Severs et al. 2013). While feminist scholarship on policy-making has argued that white men after 9/11 have increasingly claimed to act on behalf of Muslim women (Fekete 2006;Roggeband and Verloo 2007), we do not know how this plays out in parliamentary politics. Critical scholars argue that men are as heterogeneous as women and that the over-representation of men should be understood as the 'overrepresentation of privileged men' (Bjarnegård and Murray 2018: 268, emphasis original), thereby inviting intersectional analysis (Childs and Hughes 2018). As (the interests of) ethnic minority men remain invisible in the scholarship on political representation, we are, using our current methods, unable to compare ethnic minority women and men. We can, however, study the extent to which ethnic minority male MPs address issues related to the position of ethnic minority women.
We address the critique that groups are not homogenous entities in the process of substantive representation by adopting an intersectional approach, distinguishing between ethnic minority and majority women and men. While we acknowledge that these groups are by no means homogenous, distinguishing between them allows us to study who claims to act on behalf of ethnic minority women, a first step in empirically disentangling how minorities are represented. This leads us to the following hypothesis: Intersectionality hypothesis (H1). Ethnic minority women MPs are more likely than female ethnic majority, male ethnic minority and male ethnic majority MPs to actively address issues related to ethnic minority women, all else being equal.
The idea that representation is a process moves away from the idea that individual politicians decide whether or not to promote the interests of groups with which they share demographic characteristics. Political parties play a significant role in this process by assigning elected representatives to committees. While individual MPs may express their preferences to be members of specific committees, it is the party that decides (e.g. Damgaard 1995;Heath et al. 2005;Riera and Cant u 2016). In this light, the relationship between the descriptive and substantive representation of ethnic minority women may be spurious: parties may be more likely to assign female ethnic minority MPs to immigration-and social issuerelated committees, while members of these committees are more likely to address the interests of ethnic minority women. Here it is not an MP's commitment to represent her group that is driving the relationship.
Based on a study of six Latin American countries, Heath et al. (2005: 420) conclude that 'women tend to be isolated on women's issues and social issues committees and kept off power and economics/foreign affairs committees as the percentage of legislators who are women increases'. They find such exclusion more likely when party leaders or chamber presidents control committee assignments than when legislators elect one another in floor votes. In the Netherlandswhere the party has the final decision on committee assignments (Damgaard 1995)we expect ethnic minority women MPs will more likely be assigned to women and social issues committees. Cross-national research also confirms that ethnic minority MPs in many European countries, including the Netherlands, are more likely to be seated on immigration-related committees (Fernandes et al. Forthcoming). But in contrast to Heath et al.'s (2005) findings, ethnic minorities were not under-represented in power committees (e.g. economics and foreign affairs). Still, based on both studies, we expect female ethnic minority MPs to be over-represented on committees addressing minority-related affairs, meaning that the (cumulative) effects of ethnicity and gender on a MP's propensity to ask parliamentary questions on issues concerning ethnic minority women may at least partly be explained (i.e. mediated) by committee membership.
Parliamentary committee hypothesis (H2). Ethnic minority women are more likely to represent the interests of ethnic minority women due to their memberships in social issue-related parliamentary committees, all else being equal.

Data, operationalisation and methods
We rely on data on the background characteristics of individual parliamentarians collected within the framework of the Open Research Area Funded project 'Pathways to Power: The Political Representation of Citizens of Immigrant Origin in Seven European Democracies' (PATHWAYS). 2 For all the legislative periods under study, coders familiar with Dutch society and politics were asked to evaluate whether each parliamentarian could be perceived by voters as belonging to an ethnic minority group on the basis of their name and official photograph. We only included MPs in the 'identifiable minority' group if coders indicated no ambiguity; when coders used the 'ambiguous' code, the MP was categorised as not belonging to an identifiable minority. Coders also marked the country of birth and nationality at birth of MPs and their parents. For our purposes, we excluded those identified as ethnic minorities from within Europe (e.g. German, Greek) and North America, as it is the 'integration' of immigrants with so-called 'non-Western' backgrounds that has been politicised in Dutch political debate (van Kersbergen and Krouwel 2008). For the gender of the parliamentarian, we also relied on the PATHWAYS database.
To study the substantive representation of ethnic minority women we rely on written parliamentary questions addressed to the government by MPs (for a similar approach see Bird 2005;Saalfeld 2011;Saalfeld and Bischof 2013). Questions need to be submitted in written form to the President of the Chamber; the relevant member of the cabinet is then obliged to answer in writing within a three-week period. Questions for written answer are thus an effective tool for MPs to direct the attention of other parties and the government to their preferred issues. Unlike plenary debates and oral questions, there is no centralised gatekeeping by the presidium or governing parties; written questions thus likely reveal the priorities of MPs (Keh 2015;Russo and Wiberg 2010). Compared to voting and parliamentary speeches and oral questions, the party's control over the content of written questions is minimal (Rozenberg and Martin 2011). Written questions thus lend themselves well to evaluating the 'politics of presence'.
We searched all parliamentary questions for written answers from December 1994 to 2012 for keywords relating to ethnic minorities, women and ethnic minority women (see Online Appendix 1 for the search strings). The search strings were developed in three steps. First, we composed initial lists of search words based on our own knowledge of the issues, suggestions from colleagues and prior research (Boomgaarden and Vliegenthart 2007). Second, we examined the parliamentary questions these search strings yielded and added relevant words (and synonyms) that were not yet included in the search strings. Third, we examined a sample of the broader set of parliamentary questions containing the word 'woman' (vrouw in Dutch) for additional terms. At each step, we checked that search words did not result in mishits by manually reading a number of the additionally retrieved questions. The final search string on ethnic minorities (and immigrants) returned the most hits (4670 parliamentary questions) while the women search string yielded 381 questions and the ethnic minority women search string 171. We then combined these search strings to find questions that concerned both minorities and women by selecting questions either containing both a hit on the ethnic minority search string and a hit on the women search string, or a hit on the minority women search string. This yielded a total of 236 written questions posed about minority women's interests in an 18-year period (1995-2012) covering seven parliamentary sessions.

Variables and model specification
To evaluate our hypotheses, we estimate logistic regression models predicting whether a question addresses ethnic minority women's interests using characteristics of the MPs asking the question. In other words, our observations are made up of all parliamentary questions in the research period and the dependent variable is a dummy indicating whether women and ethnic minorities are mentioned in the question. We rely on a total of 35,958 parliamentary questions, 236 of which were classified as dealing with ethnic minority women's issues. While the share of events compared to non-events may be low, the number of events meets Vittinghoff and McCulloch's (2007) recommendation of at least 5-9 events per independent variable. Ordinary logistic regression is therefore appropriate, even though we verified that our findings are robust against methods for rareevent analysis.
The first independent variable is a dummy capturing whether the first petitioner of the parliamentary question can be identified as an ethnic minority. Second, we are interested in the effect of gender, which we again measure by means of a dichotomous variable, coded with a 1 if the MP is a woman and 0 if the MP is a man. To test the intersectionality hypothesis (H1), we calculated the interaction between gender and ethnicity.
Our second hypothesis builds on the idea that parties assign MPs to parliamentary committees by considering characteristics such as their ethnic background and gender. The hypothesis subsequently posits that membership in the relevant committee rather than the ethnicity and gender of the petitioner determines whether issues of concern to minority women are raised. In other words, the hypothesis suggests that (partydetermined) committee membership mediates the effect of MP background on the content of their parliamentary questions. To test this, we evaluate whether the (combined) impact of ethnicity and gender diminishes once we control for committee membership, using two different operationalisations. The first uses the PATHWAYS coding, which records whether a committee deals with policies that may potentially impact citizens of immigrant origin and their descendants. A broad range of committees is included in this list, such as Social Affairs or Housing, Infrastructure and the Environment. Our second operationalisation is more specific and only includes committees that explicitly deal with policy on the 'integration' of ethnic minorities in society. For both variables, any MP who was a member of the committee in any function was coded with 1; all other MPs were coded with 0. 3 The dataset consists of a large number of parliamentary questions asked by a much smaller number of MPs. To account for this clustering, we estimate models with robust standard errors clustered at the level of MPs. In addition, to examine whether party ideology or any other unobserved party characteristic is driving the observed effects, we use party dummies as control variables. We estimate alternative models to test the robustness of our findings in Online Appendix 3.

Context: gender and ethnicity in the Dutch parliament
The first ethnic minorities elected to the Dutch parliament were Dutch citizens from the (former) colonies. In 1933, Roestam Effendi, a man from the then Dutch East Indies, was elected for the Communist Party (Rath 1985). The next MP with roots in Indonesia was elected more than 50 years later, in 1986, this time for the Labour Party. In 1994, the number of ethnic minority MPs rose from one to eight (out of a total of 150). Since then, there has been a rapid increase in the number of ethnic minorities in the Dutch parliament. Since 2003, their percentage in parliament is comparable to their share of the Dutch population in the period under study (about 10%). 4 Although the first ethnic minority MPs in the Netherlands were men, they have been outnumbered by their female counterparts since 1998 (see Table 1). During the 2006-2010 term, 72% of ethnic minority MPs were women. Among female MPs, the share of ethnic minorities increased steadily between 1994 and 2006, from 2% to 17%. Overall, if we compare the share of women among ethnic minority MPs and the share of ethnic minorities among women MPs, ethnic minority women seem to be very well represented. In other words, if we cross the axes of gender and ethnicity for minority women, the outcome is extremely positive, whereas the outcome for ethnic minority men is negative. This is surprising since among the ethnic majority, men are much better represented in parliament than women. The success of ethnic minority women in Dutch parliamentary politics underlines the importance of analysing the representation of minority MPs intersectionally, as it suggests mechanisms other than those one would intuitively expect by examining gender or ethnicity in isolation.
Dutch colonial and immigration history is clearly reflected in the country of birth and nationality of minority women MPs and their parents, as shown in Table 2. The best represented (11 in total) in this period are colonial and postcolonial immigrants from Aruba, Curacao, Indonesia  Table 3 provides a descriptive overview of the relevant committee memberships and parliamentary chamber roles. It shows that the four gender and ethnicity combinations hardly differ in the number of committees on which they sit (all MPs sit on about four committees). Next, the table shows membership in the social affairs committee which deals with women's emancipation. Differences are not large, but women are more likely to sit on this committee (18-19% of women do), while ethnic minority men form a middle group between women and ethnic majority men. We further distinguish between two types of immigration-related committees: one restrictive set of committees that explicitly addresses the 'integration' of immigrants into society, and a second broader set that includes any committee that might deal with immigration-related issues. Both show the same pattern: ethnic minority (both male and female) MPs are much more likely to sit on these committees. For both social affairs and immigration-related committees, descriptive characteristics predict membership in the expected direction. Leadership roles in the chamber, displayed in the lower half of Table 3, are predominantly held by ethnic majority MPs (both men and women) and significantly less often by ethnic minority women. Membership of the parliamentary party leadership group shows a clear double (ethnicity and gender) disadvantage: 30% of ethnic majority men, 27% of ethnic majority women, 23% of ethnic minority men and 15% of ethnic minority women have such roles. As these numbers are percentages within the specific groups, they are already corrected for differences in their numerical presence in parliament.

Descriptive statistics
Do ethnic minority women MPs try to champion the interests of ethnic minority women? Figure 1 shows that attention for immigrant and ethnic minority women's issues in the Dutch parliament has increased over the past decades. As the total number of written questions tabled in the Dutch Lower Chamber increased over the years, the figure shows both the absolute number of questions (upper half), and the share of questions on the topic relative to the total (lower half). As the upper panel shows, only a handful of questions were asked per year between 1994 and 2001. Since then, there has been a steady increase in issue attention, albeit with variations. Questions concerning ethnic minority women are broken down by party in Online Appendix 2. Parties from both sides of the ideological aisle address the issue (for example on the right the anti-immigration parties Pim Fortuyn List and Party for Freedom and on the left GreenLeft and the Labour Party), while there are also parties on both sides that never mention minority women (for example the anti-immigration Centre Democrats and the pro-animal rights Party for Animals). It apparently makes little difference whether the petitioner's party is on the left or right. If we turn to the characteristics of the petitioners in Table 4, we see that very few MPs raise issues concerning ethnic minority women in parliamentary questions: 3% to 13% table questions as the first petitioner, while only 3% to 18% raise issues at all (as first petitioner or co-petitioner). In absolute numbers, 5 to at most 31 MPs per term ask parliamentary questions related to the position of ethnic minority women. If we look only at women MPs, we see that the percentage is hardly higher than for all MPs. If we only consider ethnic background, minority MPs on average have a higher propensity to discuss the position of minority women, with 10% to 36% asking such questions each term.
Are ethnic minority women more likely to represent ethnic minority women?
In Table 5 we present the results from the logistic regressions. Model 1 estimates the effect of being a woman and of belonging to an ethnic minority without interaction between the two. Both coefficients are positive and statistically significant. On average, the logged odds that women will table a question that addresses the interests of ethnic minority women are 0.41 higher than for men. In terms of odds ratios, women are 1.51 times more likely to ask questions than men. 6 As for the effect of ethnicity, the logged odds are 1.38 higher for visible ethnic minorities, meaning that they are 3.97 times more likely than ethnic majority MPs to address the interests of ethnic minority women. 7 In other words, female and ethnic minority MPs on average more often raise issues concerning minority women. But this only concerns the additive effects of gender and ethnicity. We proceed to evaluate how both characteristics interact. In Model 2, we include an interaction term between gender and ethnicity, allowing both characteristics to interact intersectionally. In this way, ethnic minority status can have a different effect for men and women, just as gender can have a different effect for ethnic minority and majority parliamentarians. Indeed, in line with the intersectionality hypothesis (H1), the interaction term is positive and significant, while the main effects of gender and minority status are no longer significant. How this plays out for the four groups is displayed on the left-hand side of Figure  2, which shows for each group the probability predicted by the model that a parliamentary question raised by a MP addresses the interests of ethnic minority women. For ethnic minority women MPs, the probability is 2.9%; for ethnic minority men and ethnic majority MPs, the likelihood is only 0.3% to 0.5% per question. Clearly, ethnic minority women MPs  1994-1998 1998-2002 2002-2003 2003-2006 2006-2010 2010-2012   are much more likely to ask questions concerning the interests of ethnic minority women than other members of parliament. It could be that the ideology of the petitioners' party is driving the results. That is, certain parties may be more likely to discuss as well as to attract ethnic minority women as politicians. To account for potential omitted variable bias, we include party dummies in Models 3 and 4. As Table 5 shows, the results remain largely unchanged, indicating that even when the petitioner's party is held constant, the descriptive characteristics of legislators continue to predict whether they will address the interests of ethnic minority women. Finally, in Models 5 and 6 we account for potential clustering in the standard errors, necessary because a large number of parliamentary questions are posed by a small group of MPs. This considerably increases the uncertainty around our estimates. The interaction is now only significant at the P < 0.10 level (two-tailed). So although ethnic minority women ask more questions concerning ethnic minority women, there are large individual differences in the extent to which they do so. This is also reflected in the increased confidence intervals displayed on the right-hand side of Figure 2, showing the predictions based on Model 6. Still, this figure underlines at the p < 0.05 level that policy issues concerning ethnic minority women were indeed tabled overwhelmingly by ethnic minority women MPs, suggesting a link between descriptive and substantive representation. That said, the connection is tenuous as it depends on a very small number of individuals.
For example, Khadija Arib, a female Moroccan-Dutch Labour Party parliamentarian, tabled 27% of all questions under study as the first petitioner. Arib, who served in parliament in all seven terms under study, is an exceptional politician who has shattered multiple glass ceilings. She has been a MP for more than 18 years, making her one of the longest serving parliamentarians in Dutch history. In 2016 she was appointed as the first speaker of parliament with an immigrant background. The link between descriptive and substantive representation would thus be weaker Notes: †p < 0.10; Ã p < 0.05; ÃÃ p < 0.01; ÃÃÃ p < 0.001 (two-tailed). Models 1a and 1b are used in Figure 3. without Arib's presence. Ethnic minority MPs who stick around for more than one term and break through existing barriers are exceptional fighters (see Arib 2009). It is hardly surprising that those with solid positions in the party are especially outspoken and become critical actors. Hypothesis 2 posits that the party leadership influences the issues MPs raise in parliamentary questions through their assignment to committees. In this reasoning, the party leadership sorts parliamentarians into committees depending in part on their background characteristics such as gender and ethnicity, with committee membership then largely determining which issues a MP raises in written questions. Do these roles and memberships explain who asks questions that concern the interests of ethnic minority women? We predict whether a parliamentary question does so using a range of variables capturing committee membership; the results are shown in Table 6. The prediction improves markedly, raising the pseudo-R 2 from 5% to 9%, providing support for Hypothesis 2. Figure 3 plots the predicted probability of a parliamentary question addressing ethnic minority women for the four gender and minority combinations, based on two models: one with and one without committee controls. Controlled for committee membership, ethnic minority women MPs are predicted to ask about minority women issues in 1.3% of their questions, instead of the 2.9% of questions in the model without committee controls. This effect is still statistically significant, and higher than for the other groups, which are predicted to devote around 0.1-0.5% of their questions to these issues. In other words, part but not all of the effect of being an ethnic minority woman runs through the membership of relevant committees. This indicates that part of the effect of gender and minority background is mediated by committee roles, and hints at the influence of the party leadership in steering the questioning behaviour of individual parliamentarians.

Conclusion
The descriptive representation of ethnic minority women in the Dutch parliament has been so successful that the gender gap among minorities has been either negligible or reversed in most years under study. While this mirrors patterns found in the US and other European countries, we know little about how the greater numerical presence of ethnic minority women in elected office influences the issues they raise. In this article we have studied the link between descriptive and substantive representation intersectionally to reveal the differences within and between gender and ethnic groups. Do ethnic minority women represent ethnic minority women? Our data confirm that female parliamentariansregardless of party affiliation and ethnicityare more likely than male parliamentarians to table questions related to the position of ethnic minority women. Likewise, the data confirm that ethnic minority parliamentariansregardless of gender and party affiliationare more likely to address issues related to the position of ethnic minority women than their ethnic majority colleagues. But in both cases, the interests of ethnic minority women are not addressed by just any woman or ethnic minority MP, but by ethnic minority women parliamentarians. This intersectional link is confirmed in H1. That said, we find the proportion of ethnic minority women MPs who table parliamentary questions concerning the interests of ethnic minority women to be very small. Substantive representation based on the descriptive representation of ethnic minority women is thus fragile as it depends on a small number of individual 'critical actors'.
With H2 we examine whether ethnic minority women MPs are more likely to represent the interests of ethnic minority women due to their membership in social issue-related parliamentary committees. Committee membership indeed explains why ethnic minority women table more parliamentary questions on policy issues related to ethnic minority women. When we control for committee membership, ethnic minority women MPs still table more questions regarding minority women than other parliamentarians, meaning that our Intersectionality hypothesis (H1) stands.
Finally, the data show a remarkable shift over time. There was hardly any attention to immigrant and ethnic minority women before 2001. The politicisation of Islam and immigration in the wake of 9/11 alongside the rise of rightist parties in parliament spurred the interest in the position of ethnic minority women. In this same period, ethnic minority women MPs began to outnumber their male colleagues and took active part in the debate.
These findings contribute to our understanding of the political representation of structurally under-represented groups and underline the value of an intersectional approach to the study of representation, which yields more refined results than approaches that consider gender or ethnicity in isolation. It is the specific intersection between being a woman and belonging to a visible ethnic minority that appears to be an important indicator for the substantive representation of ethnic minority women. This effect would be missed in a study that focused on either gender or ethnicity in isolation.
Our study raises a number of questions for further research. First, we have focused on parliamentary questions as a proxy for substantive representation. While this is a common approach in legislative studies, it is a narrow understanding of substantive representation. The next step is to grasp what kind of policies have been championed for ethnic minority women and the extent to which these have been differently influenced by ethnic minority women parliamentarians and their colleagues. Second, our research shows that the link between descriptive and substantive representation is fragile and depends on a small number of MPs. Crossnational analysis should reveal if this also holds for other countries. What factors explain why some ethnic minority women are so outspoken while others are not? In the Dutch case, the key critical actor was an incumbent who, despite substantial push-back, over the years cemented her position in the party. But the question remains how critical actors become critical; is it facilitated by 'critical mass'? Finally, the link between the descriptive and substantive representation of ethnic minority men merits attention in studies of group representation. To what extent do they push behind the scenes for the recognition of ethnic minority men's rights? These questions require quantitative and qualitative comparative study to capture both the visible and formal as well as the invisible and informal aspects of representation. Notes a script provided by Carsten Schwemmer, which was updated by Mats Willemsen and Erik de Vries; Takeo David Hymans edited the manuscript.

Disclosure statement
No potential conflict of interest was reported by the authors.