How do minoritized migrant-background citizens perceive their political representation? An intracategorical perspective

ABSTRACT

Although our insights into numerical political representation are expanding, we know remarkably little about whether minoritized migrant-background citizens feel represented as such. This study addresses to what extent there are intersectional differences in perceived representation among minoritized migrant-background citizens. We synthesize two separate literatures to develop an overarching boundaries-privilege framework, which we apply to the Dutch Ethnic Minority Election Study 2021. This shows for instance, that those with university degrees and those who identify as Muslim perceive lower political representation. This provides particular support for the idea that these perceptions are shaped negatively by exclusions grounded in discursive boundaries.

KEYWORDS:

• political representation
• perceived political inclusion
• ethnic minorities
• Muslims
• boundaries
• privilege

Introduction

Political representation has major democratic significance, so it is not surprising that numerous works have studied the representation of marginalized groups, including minoritized migrant-background citizens¹ (Bird, Saalfeld, and Wust 2011; Mügge, van der Pas, and van de Wardt 2019; Sobolewska 2013). However, the empirical study of subjective perceptions of representation is rather new (De Jong 2024; De Mulder 2022; Holmsberg 2020), and remarkably little is known about whether minoritized migrant-background citizens consider their group and interests to be represented sufficiently, even though group-based differences in perceived representation are deemed “normatively unwelcome” (Holmsberg 2020, 423). How well do varying minoritized migrant-background citizens feel represented?

This diversity among minoritized people is often lost in comparative studies, which tend to focus on differences in trust or participation between majority and minority citizens (Spierings and Vermeulen 2023; Verba, Burns, and Schlozman 2003), which tends to homogenize both groups. Another strand of literature zooms in on very specific groups qualitatively (De Jong 2024; De Jong and Mügge 2023; Murray 2016), which sheds little light on how experiences differ across minoritized groups. The present study’s first and main contribution to the literature is thus to examine what dividing lines exist among migrant-background citizens in their perceptions of descriptive and substantive political representation of migrant-background citizens. By taking this intracategorical approach, we focus on “particular social groups at neglected points of intersection” (McCall 2005, 1774)-particularly intersections with socio-economic status, gender, ethnic background and religion-which promises to refine our thinking on the perceived representation of minoritized people by revealing the complexity within this group (Geurts 2022).

We also contribute to the literature on migration and representation empirically, by using unique data on perceived representation among minoritized migrant-background citizens that make up around 10 per cent of the Dutch electorate: the Dutch Ethnic Minority Election Study (Lubbers, Sipma, and Spierings 2021) for the 2021 elections. These data allow us to explore how minoritized migrant-background citizens perceive political representation, rather than focusing on objective descriptive representation (see also De Mulder 2022; Holmsberg 2020), as other studies have done (Lowande, Ritchie, and Lauterbach 2019; Mügge 2016; Sobolewska 2013). In particular, these data add to the literature because they question respondents’ perceptions of the political representation of migrant-background citizens specifically. This allows us to empirically tackle the common assumption in the literature that minoritized migrant-background citizens perceive little representation because of their ethnicized identity (Bird, Saalfeld, and Wust 2011).

Although these data limit us to 300 minoritized migrant-background citizens in the Netherlands only, the Dutch 2021 elections present an intriguing case. On the one hand, the Netherlands seems comparable to other Western European countries in terms of xenophobic public sentiments and increasing descriptive representation of minoritized people in recent decades. Nowadays, certain, but not all, migrant-background groups are even overrepresented in Dutch politics, which means we can go beyond assessing who perceives their objective underrepresentation (e.g. De Vroome, Verkuyten, and Martinovic 2014; Murray 2016; Verba, Burns, and Schlozman 2003) to exploring who perceives little representation in the face of legion parliamentarians with a migration background. Simultaneously, the 2021 election was a specific “Covid election” that took place while political trust plummeted among general publics. Especially as the elections were triggered by institutional racism and focused on the (ethnicized) policy responses to the Covid pandemic, minoritized people across the board might have had little warm feelings toward politics. Consequently, if we uncover diversity among minoritized people even in this context, we might expect it to definitely exist elsewhere.

Our third contribution is to transcend disciplinary differences; we synthesize two separate literatures to develop an overarching boundaries-privilege framework on minoritized migrant-background citizens’ perceived political representation. First, from the literature on migration and belonging, we derive arguments that people who experience more boundaries (e.g. discrimination) experience more political unwelcomeness and non-belonging, manifesting itself in lower perceived representation, among other things (Alba 2005; Eijberts, Ghorashi, and Rouvoet 2017; Ghorashi 2018; Simonsen 2018). Second, and inversely, from the general political trust and external efficacy literature we derive arguments that more privileged people are more satisfied with the political status quo, manifesting itself in more positive evaluations of the system and this also higher perceived representation (Newton 2009, 357; also Spierings and Vermeulen 2023; Whiteley 1999). Applying this framework, we derive and test hypotheses on the role of socio-economic status, gender, religion and migration history, whereby the privilege logic and the boundaries logic lead to contradictory expectations for most factors.

Theoretical background

After assessing whether political inclusion and engagement gaps exist between ethnic minorities and ethnic majorities and assessing which differences between these groups explains such differences, scholars turned their attention to processes specific to migration and belonging to a minority group (see Bird, Saalfeld, and Wust 2011). This line of research moves away from the intergroup comparison between majorities and minoritized migrant-background citizens, and doing so it allows for a more open perspective than simply copy-pasting majority-based insights onto minoritized people. The current study uses such an intracategorical approach (see also De Jong and Mügge 2023; Geurts 2022; McCall 2005).

Perceived representation

A rich literature on (women’s) objective representation in politics has differentiated between different dimensions of representation, whereby the core focus in current debates concerning minoritized migrant-background citizens’ representation is on descriptive and substantive representation: are minoritized migrant-background citizens partaking in politics as politicians and are their interests part of the political decision-making process (Pitkin 1967).

Accordingly, a large literature has focused on explaining levels of descriptive and substantive representation (see Wängnerud 2009), also of ethnic minority groups (e.g. Lowande, Ritchie, and Lauterbach 2019; Sobolewska, McKee, and Campbell 2018). This, however, mainly refers to what has been labeled “objective” representation (De Mulder 2022), i.e. the supply side of representation: the degree to which politicians reflect minoritized migrant-background people’s demography and interests (see Holmsberg 2020).

Our focus, however, is on what De Mulder (2022) labeled “perceived subjective representation” (see also Holmsberg 2020), which can be simply understood as the degree to which minoritized migrant-background citizens perceive they (in this study: as a group) are represented. This topic has been given little empirical attention (Holmsberg 2020) and is very much undertheorized (De Mulder 2022), with existing studies focusing fully on the perception of substantive representation (De Mulder 2022; Holmsberg 2020). We build on and expand this literature by (a) focusing on the perceived representation of migrant-background citizens, (b) in (perceived) substantive and descriptive representation, and (c) by presenting a theoretical framework to understand the more general roots of differences in perceived representation.

Regarding our focus on different dimensions of perceived representation, the few existing studies looking into this (as well as our data – see Section 3.2) show these are empirically distinct, but also that they clearly correlate. Minoritized migrant-background citizens discussing both dimensions value both and for instance stress the importance of shared histories with politicians which directly relates to both (Dagevos et al. 2024). And De Mulder (2022) recently distinguished dimensions of perceived substantive representation in terms of different acts (e.g. listening, knowing, acting), but in terms of explaining vote choice the dimension showed similar results. In terms of perceptions of descriptive and substantive representation, we therefore expect that these are strongly interrelated and probably more strongly so than their objective counterparts (from which this distinction draws). Perceptions of both descriptive and substantive representation tap into more general experiences and given the state of the explanatory literature of perceived representation (see Holmsberg [2020]’s overview chapter) it is far from evident how the intracategorical differences in perceived representation might differ between the dimensions. Considering this all, our theorization below concerns “perceived political representation” in general, while empirically assessing these relationships for descriptive and substantive representation separately too. In the conclusion, we will reflect on the (limited number of) differences found.

Moreover, we focus specifically on the perceived representation by and of migrant-background citizens and their interests. In Western Europe and the Netherlands specifically, the topic of migration is politically salient and citizens with a migration background from Asia, Africa, or South America are minoritized along ethnic and religious lines, and in this politicization and public discourse migration history, ethnicity and religion are fluidly interwoven. In line with this context, minoritized migrant-background citizens tend to identify with different groups and at different conceptual levels, of which one is “migrant-background people”, at least to some extent (Dagevos et al. 2024). Therefore, in the theorize below we assume that many minoritized migrant-background citizens consider questions about people with a migration background to also be about themselves. In the empirical part of this study, we explore these interlinkages in more detail.

An explanatory understanding of perceived representation: privilege and boundaries

With the exception of some mechanism for singular explanatory variables, particularly by Holmsberg (2020), there are basically no explanatory theoretical framework for understanding variation in minoritized migrant-background citizens’ perceived representation. Evidently, there is a very rich literature on the demand side of representation, which presents concepts like representative claims (Celis et al. 2014; Saward 2006) and the subject of representation (i.e. who can represent what group [Celis and Childs 2008; Saward 2008]). And that literature touches upon responsiveness and feeling of representation with implicit assumptions that systems with groups being visibly present, with a more responsive process and with more responsive outcomes leads to stronger perceived representation. However, it lacks a clear theoretical framework on how perceived representation comes about.

The literature on perceived or subjective representation is only just budding (see previous section), particularly regarding explanatory work. On the latter, particularly Holmsberg (2020) weighs in, but only points to some specific factors and macro-level mechanisms. Given our focus on understanding intracategorical differences among minoritized citizens with a migration background, we need to zoom out to formulate a more general framework (in which these specific factors can be integrated, as will be done below).

We develop a framework along the lines of privilege and boundaries, which draws from the general political trust and external efficacy literature (Newton, Stolle, and Zmerli 2018; Whiteley 1999), also applied to migration (Spierings 2019; Spierings and Vermeulen 2023) and the migration and ethnic studies literature on belonging (Eijberts, Ghorashi, and Rouvoet 2017; Geurts 2022; Ghorashi 2018; Simonsen 2018). Each of these two literatures presents its own core logic, which revolves around understanding how societal group-based differences come into being. Below, per literature, we will briefly unpack these core logics and translate them to perceived represented. Next, we apply these logics to specific expectations on SES, gender, migration history, and religiosity. In most cases, the two mechanisms lead to the contrasting expectations on the specific factors.

The literature on political trust and external efficacy focus on concepts akin to perceived representation. Political trust is grounded in a more basic disposition that others (e.g. politicians) will behave in such a way that they prevent to harm others. Empirically, this refers to questions whether people trust politicians, parliament, government and political parties. External efficacy more particularly refers to whether people feel that they can effectively influence politics via politicians, phrased in questions like whether “people like me” have an influence on politics and MPs are interested in one’s opinion or votes. Moreover, trust and efficacy are part of the larger concept “evaluation the political system”, (Spierings and Vermeulen 2023).

One of the core mechanisms, presented in the literature of trust and efficacy, is that people are more likely the evaluate aspects of the political system positively if they benefit from it or, phrased some more leniently, do not suffer from it. More concretely, it is argued that if one belongs to a group that is more privileged in society, they experience more (access) to resources, more freedoms, and less troubles. Belonging to a more privileged groups is consequently linked to being more satisfied with the political elite and system (e.g. Newton 2009; Newton, Stolle, and Zmerli 2018; Spierings and Vermeulen 2023; Whiteley 1999). People who feel they have succeeded in life – who are society’s “winners” – simply have (unconsciously) less reason to judge the system negatively. Thus: those who are less successful or less privileged are likely to partly blame the system for creating hurdles and holding them back. In brief: “Trust seems to be a privilege of the rich, successful, and educated” (Newton 2009, 357).

We argue that these arguments can be extended to minoritized migrant-background citizens’ perceptions of the representation of migrant-background citizens. Judging perceived representation of one’s group can be considered a manifestation of an underlying satisfaction with the functioning of the democratic system, particularly in terms of group representation. However, also among minoritized migrant-background citizens, relatively privileged subgroups might experience the system and its manifestations more positively, which is in turn not unlikely to be reflected in their judgement on the representation of migrant-background citizens. By theorizing this from an intracategorical perspective (i.e. zooming in on diversity among minoritized migrant-background citizens; not comparing them to their majority counterparts), we allow for hypotheses that acknowledge that hierarchies dominant in society at large might be absent or inverse among minoritized migrant-background citizens. How the privilege perspective translates to specific hypotheses is theorized after discussing the literature on belonging and boundaries directly below.

The literature on the belonging of minoritized migrant-background citizens also focuses on societal experiences and how these are related to one’s minoritized migrant background. Belonging basically refers to the feeling of being part of a group or larger entity (Geurts 2022; Geurts, Davids, and Spierings 2021), in this literature often the society of living after migration. That society is also the object of politics: are people considered part of the populace that is being represented? Clearly, belonging and perceived representation share a root in whether minoritized migrant-background citizens feel part of the polity.

Whereas the privilege literature take a more materialist approach, the belonging literature draws more strongly on a constructivist approach and puts the spotlight on the influence of constructing social boundaries (Alba 2005; Geurts, Davids, and Spierings 2021; Simonsen 2018). In general it is assumed that if one is at the outsider side of the discursive boundary, this translates to less belonging (e.g. Geurts, Lubbers, and Spierings 2020) and, we argue consequently to less perceived representation. Again and importantly, multiple and intersecting boundaries can exist simultaneously and minoritized migrant-background citizens experience boundaries not purely or simply based on their migration background. This then brings us to the question how this such boundary-drawing logic translates to intracategorical differences in representation?

From the belonging literature, we know that social positions shape the degree to which one perceives discrimination, which in turn shapes people’s sense of belonging. Specifically, the integration paradox literature finds that those with a higher socio-economic position perceive more exclusion in society resulting in a lower sense of belonging (Geurts 2022). The notion of boundaries thus focusses on perceptions, which are grounded not just in (material) positions but also in expectations and specific experiences that create hurdles to meet those expectations. This causal logic running from social positions (e.g. education, gender, migration history) to experiences of discrimination and belonging is likely to apply to perceived representation too. People in social positions that experience or are aware of present boundary making, discrimination and exclusion are more likely to perceive political representation more negatively.

Hypothesizing differences in perceived representation

From the privilege-boundaries framework, we can derive contrasting expectations regarding socio-economic status (SES), particularly education and occupational status. As these elements are core ingredients for success, the privilege logic expects that more theoretical (“higher”) educated² minoritized migrant-background citizens and those with a higher occupational status are more likely to experience the system more positively, of which indications have been found among the general population in descriptive statistics (Holmsberg 2020, 424,cf.427).³ Therefore, we expect that:

H1a: Higher SES (theoretically educated; higher occupational status) minoritized migrant-background citizens perceive more political representation of minoritized migrant-background citizens than lower SES (more practically educated; lower occupational status) minoritized migrant-background citizens.

Simultaneously, minoritized migrant-background citizens with a higher SES might be more sensitive to subtle forms of discrimination, as they fulfilled the requirements of the receiving society but are still treated discriminatorily (Geurts, Lubbers, and Spierings 2020; 2021). Moreover, theoretically educated minoritized migrant-background citizens might run into discrimination more and have more political knowledge, including on the marginalization of minoritized representatives (Erikson and Verge 2022; Mügge 2016). Therefore, higher SES persons might experience barriers to representation more strongly, which, following the boundaries logic, translates to a stronger dissatisfaction with the current system and representation.

H1b: Higher SES (theoretically educated; higher occupational status) minoritized migrant-background citizens perceive less political representation of minoritized migrant-background citizens than lower SES (more basically educated; lower occupational status) minoritized migrant-background citizens do.

Privilege might similarly shape gender differences–even though gender gaps do not figure extensively in the general trust and external efficacy literature (Uslaner 2018) and descriptive statistics show only a slightly stronger feeling of representation among men (Holmsberg 2020, 424,cf.426). Gender privilege in politics among minoritized migrant-background communities is a complex issue, however.

On the one hand, while women are underrepresented in Dutch parliament (fifty-seven female versus ninety-three male parliamentarians), among minoritized migrant-background parliamentarians, women are overrepresented (Mügge, Runderkamp, and Spierings 2024). While this also reflects strategic choices of political parties to signal diversity (“two birds with one stone”), minoritized migrant-background women might see themselves represented in politics more so than men. A privilege perspective thus implies that:

H2a: Minoritized migrant-background women perceive more political representation of minoritized migrant-background citizens than minoritized migrant-background men.

Still, in public debates, minoritized migrant-background men are considered to be more privileged in terms of income and community status. Although minoritized women are evidently not the meek sheep they are sometimes depicted to be in political discourse, this discourse and conservative group norms on women’s political roles might prevent women from making use of communal organizations with bridging ties to other political organizations (Michon and Vermeulen 2013; Tillie 2004). As these organizations are a major communication channel, minoritized migrant-background women might perceive less representation. Such mechanisms may be reinforced by sexism in society and in media discourses that “paint a picture of women and ethnic minority politicians as different and out of place” (Runderkamp et al. 2022, 567; see also Haraldsson and Wängnerud 2019), resulting in lower perceptions of representation. In this case, privilege and boundaries might thus reinforce each other, so that minoritized migrant-background men might feel better represented.

H2b: Minoritized migrant-background women perceive less political representation of minoritized migrant-background citizens than minoritized migrant-background men.

Moving to ethnic differences, in the Netherlands, boundaries exclude groups that are well-represented numerically, meaning that partly contradictory expectations can be formulated. Moroccan and Turkish background citizens have been numerically overrepresented in Dutch parliament for years (Mügge, Runderkamp, and Spierings 2024; Otjes 2022). While these communities have been most excluded in terms of boundaries drawn in political discourse, they have organized themselves politically best, and are now more visibly present. For instance, DENK has put forward party leaders with such backgrounds. These MPs, also including other Turkish- and Moroccan-Dutch MPs like Bouchallikht from the Greens (first MP wearing a headscarf), stand out within the exclusionary political discourse. Consequently, their visibility and political representation increases. In terms of actual presence (although not treatment), they are privileged compared to other ethnic groups.

H3a: Minoritized migrant-background citizens with a Moroccan or Turkish background perceive more political representation of minoritized migrant-background citizens than minoritized migrant-background citizens with a different background.

A boundary-drawing logic, however, argues that these communities are excluded via ethno-religious fault lines. In Western European public debates, “migrant” and “Muslim” are regularly used interchangeably, as radical right politics has put Islam more central since the 2000s (De Lange and Mügge 2015). This shows how certain characteristics ascribed to some minoritized migrant-background citizens are problematized or politicized while others are not. In Dutch debates, “Moroccans” feature among anti-migration slang relatively frequently while other groups do not. Albeit to different degrees, Caribbean, Middle Eastern and (North) African (MENA), and Turkish Dutch citizens are discursively blended into non-White luck seekers, who are culturally different and threatening. For East and Southeast Asia and Latin American background citizens, this discourse is far less clear, even though all these groups are considered “non-Western”. People on the “negative” side of social boundaries might perceive minoritized migrant-background citizens’ representation more negatively. Or, put differently, people for whom the discrepancy between discourse and lived experience of effort to belong is greatest might experience minoritized migrant-background representation most negatively. This all supports the idea that particularly Asian-Dutch and Latin American-Dutch people experience relatively greater representation.

H3b: Minoritized migrant-background citizens with a Latin American or (non-MENA) Asian background perceive more political representation of minoritized migrant-background citizens than minoritized migrant-background citizens with a different background.

This boundary-drawing logic also leads us to expect that citizens who consider themselves Muslim and more strongly so are more likely to feel maltreated (Sobolewska, McKee, and Campbell 2018). Part of the strong anti-Muslim discourse is the lack of a clear and similarly univocal countervoice. The above-mentioned party DENK does refer to Islam and Muslims in its platform, but it does not explicitly present itself as a religious party. Moreover, up until the 2021 elections, no MP affiliated visually and explicitly with Islam. This might feed into Muslim minoritized migrant-background citizens’ assessment of the representation of minoritized migrant-background citizens in general, because from their perspective it is more likely to conclude the group (in all its diversity) is ill-represented. As a privilege argument cannot be made for Dutch Muslims, boundaries lead us to expect that:

H4a: Minoritized migrant-background citizens who consider themselves Muslim perceive less political representation of minoritized migrant-background citizens than minoritized migrant-background citizens who do not consider themselves Muslim.

H4b: More religious Muslim citizens with a migrant-background perceive less political representation than less religious Muslim citizens with a migrant-background citizens.

Lastly, the boundary literature often draws attention to the distinction between “first- and second-generation” migrants: people born abroad versus in the Netherlands. Contrary to expectations, there is limited evidence that the descendants of migrants experience less discrimination and greater belonging than migrants (De Vroome, Verkuyten, and Martinovic 2014; Maliepaard, Lubbers, and Gijsberts 2010). Descendants of migrants still perceive exclusions by ethnic boundaries while they are often integrated in economic and cultural terms, particularly compared to their parents. For the children of migrants, the discrepancy between effort and reward is much larger and consequently, they feel heard less, which for example is illustrated by research studying differences in political trust in institutions (Röder and Mühlau 2012). Again, the privilege argument can hardly be made for different “generations”–public discourse hardly discusses whether politicians are so-called first or second generation–and therefore we here also only expect differences along the boundary argument. Simultaneously, the boundary logic links to the subjective representation literature, which suggests that older democracies allow for more experience according to social and institutional learning mechanisms (e.g. Holmsberg 2020, 414, 421, cf.426). This might mean that descendants of migrants, who grew up in their democracy, are more knowledgeable about its workings and have more accurate expectations, even if grim. While this nuance can dampen the boundary effect, our hypothesized starting point is that:

H5: The descendants of minoritized migrants perceive less political representation of minoritized migrant-background citizens than minoritized migrants.

Data and methods

Data

We use the Dutch Ethnic Minority Election Study (DEMES) 2021 (Lubbers, Sipma, and Spierings 2021). Statistics Netherlands drewthe original random sample based on registration categories: was drawn among first- and second-generation migrant-background citizens with a “non-Western” origin (including African, Asian (excl. Indonesia and Japan) and Latin-American countries) who hold Dutch citizenship. Online and paper pre- and post-elections surveys were fielded. 736 respondents participated in this pre-election questionnaire (22 per cent response rate; Sipma, Lubbers, and Spierings 2021). Of them 404 also participated in the post-election survey. We include questions from both the pre- and post-election study.

The 2021 parliamentary elections were triggered by the government resigning (15/1/2021) over the (failure to deal with) scandalous social assistance anti-fraud policies, including institutional racism against ethnic minority citizens (i.e. ethnic profiling of “potential” fraudsters) (Sipma et al. 2024; Van Holsteyn and Irwin 2022). Both the social assistance affair and Covid policy responses disproportionally impacting ethnic minorities and people with a migration background (i.e. travel restrictions and ethnic minorities showing higher death rates) (Walter and Glas 2024), but these issue were marginalized during the lead up to the elections. The (policy responses on the) Covid pandemic, climate policies and the issue of “new style political leadership” became more salient (Otjes and Voerman 2022; Sipma et al. 2024). In other words, people with a migration background had the more reason to experience a lack of proper representation, but the contextual conditions might actually partly disadvantages some of our expectations as these general circumstances refer to migrant-background citizens of different genders and religions, and actually particularly Dutch citizens with an East Asian-background experienced specific Covid-related discrimination on the streets (see Odag and Moskovits 2024).

Perceived political representation of migrants

As discussed above, our dependent variable perceived political representation of migrants is based on perceived descriptive and substantive representation of citizens with a migration background. We include a mean score on the following two questions: “do you think the number of members of parliament with a migration background is too high or too low?” and “do you think the interests of citizens with a migration background are represented too little or too much?” The answer categories range from 1 “way too low/little” to 5 “way too high/much”. Given rather similar means and standard deviations, respondents could be included if they had given a valid answer on at least one of the two questions. A lower score indicates more perceived underrepresentation.

The items for descriptive and substantive representation are very strongly correlated (r = 0.84). These levels of descriptive and substantive representation of citizens with a migration background are perceived more or less the same, which validates the use of one measure on perceived political representation of migrants based on both survey questions. Yet, we also test our hypotheses for perceived descriptive and perceived substantive representation separately (see Table 2).

Relatively many respondents reported not knowing how to answer the two survey questions.⁴ We therefore employ robustness analyses including respondents who answered “do not know” on descriptive and/or substantive representation as “neither too low/little, nor too high/much”. Also, as the answered were skewed toward (way) too little representation, additional analyses are performed comparing respondents who consider representation “way (too) little” (0) with those who consider it to be “neither too little, nor too much” and “(way) too much” (1) and with those missing scores on both dependent variables (2). Descriptive statistics of all variables are in Table 1.

Independent variables

Education is measured using respondents’ highest attained educational level. We distinguish between “primary education and secondary vocational”, “tertiary vocational and higher secondary” and “university (incl. university of applied sciences)”, which ranges from less to more theoretical education. 7.2 per cent of the respondents have missing values on education (see Table A2), and were included as a separate category in our analyses. Self-identified gender includes “men” (reference category) and “women”.⁵

Occupational status is based on respondents’ self-assessed occupational class and main daily activity. We distinguish between “lower occupational status”, “middle occupational status”, “higher occupational status” and “other profession”. Table A3 in the appendix shows the detailed recoding. In line with Traber et al.’s (2022), we included unemployed people in the low-status citizens. “Inactive” includes those without a current profession while not mentioning to be unemployed (e.g. students, housekeepers, pensioners, disabled people). Respondents with a missing value (7.3 per cent) are also included as a separate category.

Respondents’ ethnic background is based on their own and their parents’ country of birth. When the parents were born in different countries, respondents were asked with which country they identified strongest. We differentiate between a background from Suriname, Dutch Caribbean, Morocco, Turkey, Sub-Saharan Africa, other MENA (Middle East and North African) countries, East Asia and other Latin America.

Religious denomination is measured by whether respondents consider themselves a member of a particular church or religious community, distinguishing between those without a religious membership, Muslims and another religions (almost exclusively Christians). Within the religious groups, we distinguish between respondents who agreed and who did not agree (incl. neither agree nor disagree) with the statement that “religion is an important part of myself”. Respondents with missing information on denomination are included as a separate category.

Migrant generation refers to whether respondents are born outside the Netherlands (“first-generation migrants”) or in the Netherlands (“second-generation migrants”).

Control variables

We include age, political trust and political interest as control variables. Age is measured in years at the time of survey. Political trust is based on three items (Cronbach’s alpha = 0.88) measuring perceived trust in government, national parliament and political parties (from 0 “no trust at all” to 3 “very high”). To measure political interest, respondents were asked how interested there are in politics from 0 “not interested at all” to 3 “very interested”.

We also control for whether respondents voted for a political party that became part of the coalition government after the election⁶ as this may affect experienced representation. Respondents who did not vote or who had a missing value (e.g. because they were not eligible to vote or did not remember whether or what they voted) are included separately.

Analytical strategy

To test our hypotheses (Table 2), we employ linear regression analyses. First, we test all our socio-demographic independent variables separately, as is summarizedin one column in Table 2 (N = 286; Model 1). Second, we include all our independent variables together, also including all control variables (N = 286; Model 2). Third, we estimated the effects for perceived descriptive representation (N = 240; Model 3) and perceived substantive representation (N = 270; Model 4) separately.

In the online appendices we present the robustness analyses, as mentioned above. First, we including respondents who answered “do not know” on descriptive and/or substantive representation as “too little, nor too much” (Table A4). This adds thirty-nine respondents compared to our main models. Second, we ran a multinomial regression analysis on perceived political representation with the categories: “(way) too little” (1); “too little, nor too much” and “(way) too much” (2);“invalid on both questions” (3) (Table A5). Third, because ethnic background and denomination are correlated, we also estimate a multivariate model without ethnic background and a model without denomination (Table A6). Altogether, the robustness analyses show very similar results; in the results sections we only discuss these results if they lead to substantially different insights.

Table 1. Descriptive statistics.

Variable	N	Mean	Std. dev	Min	Max
Perceived political representation of migrants	286	2.12	0.77	1.00	4.50
Perceived descriptive representation	240	2.09	0.76	1.00	4.00
Perceived substantive representation	274	2.13	0.85	1.00	5.00
Education
Primary and secondary vocational	44	0.15		0.00	1.00
Tertiary vocational and higher secondary	94	0.33		0.00	1.00
University	135	0.47		0.00	1.00
Missing values	13	0.05		0.00	1.00
Occupational status
Low status	19	0.07		0.00	1.00
Middle status	75	0.26		0.00	1.00
High status	69	0.24		0.00	1.00
Other occupation	41	0.14		0.00	1.00
Inactive	62	0.22		0.00	1.00
Missing values	20	0.07		0.00	1.00
Gender (men = ref.)	286	0.52	0.50	0.00	1.00
Migration background
Suriname	61	0.21		0.00	1.00
Sub-Sahara Africa	23	0.08		0.00	1.00
Former Dutch Antilles and Aruba	18	0.06		0.00	1.00
Morocco	37	0.13		0.00	1.00
Turkey	60	0.21		0.00	1.00
Other MENA	43	0.15		0.00	1.00
East-Asia	30	0.10		0.00	1.00
Other Latin America	14	0.05		0.00	1.00
Denomination
Non-religious	75	0.26		0.00	1.00
Less religious Muslims	36	0.13		0.00	1.00
More religious Muslims	52	0.18		0.00	1.00
Less religious people with another religion	78	0.27		0.00	1.00
More religious people with another religion	25	0.09		0.00	1.00
Missing values	20	0.07		0.00	1.00
Generation (first-generation = ref.)	286	0.58	0.49	0.00	1.00
Age	286	38.98	15.62	18.00	87.00
Political trust	286	1.44	0.67	0.00	3.00
Political interest	286	1.83	0.77	0.00	3.00
Vote in previous election
Coalition party	63	0.22		0.00	1.00
Opposition party	128	0.45		0.00	1.00
Did not vote	25	0.09		0.00	1.00
Missing values	70	0.24		0.00	1.00

Source: DEMES 2021, N of used sample is 300, N of original sample is 404.

Results: intracategorical dividing lines in perceived political representation of migrants

First, our results show that people with a more theoretical education, especially a university degree, perceive significantly lower political representation of migrants (see Models 1 through 4 in Table 2).⁷ This refutes the privilege logic of H1a in favor of the boundaries logic of H1b. It seems that their more privileged position does not lead minoritized migrant-background citizens with university degrees to evaluate the system more positively. Rather, these citizens might be more aware of subtle forms of discrimination and political exclusion (Geurts, Davids, and Spierings 2021), diminishing their perceived representation. Similarly, people with a middle or high occupational status perceive lower levels of political representation of migrants than people with a lower occupational status. This difference is driven by perceived substantive representation (Model 4, Table 2), as it seems to be absent for our measure on descriptive representation (Model 3, Table 2). If anything this is mostly in line with H1b; we return tho this result in the conclusion.

Table 2. Linear regression analysis on perceived political representation of migrants.

	Model 1^1			Model 2^1			Model 3^2			Model 4^3
	Bivariate models			Multivariate model			DV: descriptive representation			DV: Substantive representation
	B		s.e.	B		s.e.	B		s.e.	B		s.e.
Education
Primary and secondary vocational	ref.			ref.			ref.			ref.
Tertiary vocational and higher secondary	−0.25		0.14	−0.19		0.13	−0.19		0.15	−0.19		0.15
University	−0.31	*	0.13	−0.32	*	0.14	−0.33	*	0.15	−0.33	*	0.15
Missing values	0.12		0.24	0.16		0.24	0.05		0.25	0.29		0.27
Occupational status
Low status	ref.			ref.			ref.			ref.
Middle status	−0.71	***	0.19	−0.44	*	0.19	−0.04		0.20	−0.63	**	0.21
High status	−0.69	***	0.20	−0.48	*	0.19	−0.06		0.21	−0.68	**	0.22
Other occupation	−0.72	***	0.21	−0.39	+	0.20	−0.01		0.23	−0.52	*	0.23
Inactive	−0.59	**	0.20	−0.46	*	0.19	−0.09		0.21	−0.56	*	0.22
Missing values	−0.46		0.24	−0.19		0.23	0.38		0.26	−0.35		0.26
Gender (men = ref.)	0.00		0.09	0.03		0.09	−0.07		0.10	0.11		0.10
Migration background
Suriname	ref.			ref.			ref.			ref.
Sub-Sahara Africa	−0.10		0.18	−0.06		0.17	−0.08		0.18	−0.07		0.19
Dutch Caribbean	0.63	**	0.20	0.64	**	0.20	0.60	**	0.20	0.61	**	0.23
Morocco	−0.06		0.16	0.45	*	0.17	0.52	**	0.19	0.43	*	0.19
Turkey	0.40	**	0.14	0.80	***	0.16	0.83	***	0.17	0.81	***	0.18
Other MENA	0.35	*	0.15	0.49	**	0.16	0.43	*	0.17	0.52	**	0.18
East-Asia	0.42	*	0.17	0.39	**	0.16	0.36	*	0.17	0.38	*	0.19
Other Latin America	0.28		0.22	0.25		0.22	0.04		0.27	0.32	*	0.27
Denomination
Non-religious	ref.			ref.			ref.			ref.
Less religious Muslims	−0.23		0.15	−0.55	***	0.16	−0.49	**	0.18	−0.59	**	0.18
More religious Muslims	−0.44	**	0.14	−0.61	***	0.15	−0.40	*	0.17	−0.69	***	0.16
Less religious people with another religion	−0.02		0.12	−0.01		0.12	0.07		0.13	0.02		0.14
More religious people with another religion	−0.10		0.18	−0.17		0.17	−0.17		0.18	−0.22		0.19
Missing values	−0.26		0.19	−0.53	**	0.19	−0.47	*	0.20	−0.63	**	0.22
Generation (first-generation = ref.)	−0.19	*	0.09	0.15		0.11	0.28	*	0.13	0.12		0.12
Age	0.01	*	0.00	0.01		0.00	0.01	**	0.00	0.01		0.00
Political trust	0.26	***	0.07	0.18	**	0.07	0.14	+	0.07	0.17	*	0.08
Political interest	−0.04		0.06	−0.09		0.06	−0.12	+	0.06	−0.11		0.07
Vote in previous election
Coalition party	ref.			ref.			ref.			ref.
Opposition party	−0.27	*	0.12	−0.10		0.11	−0.13		0.12	−0.06		0.13
Did not vote	0.19		0.18	0.33		0.18	0.59	**	0.20	0.18		0.21
Missing values	−0.28	*	0.13	−0.12		0.14	−0.11		0.15	−0.17		0.16
Intercept				2.14	***	0.34	1.58	***	0.36	2.40	***	0.39
Adjusted R2				0.22			0.20			0.21

***p < 0.001, **p < 0.01, *p < 0.05, + p < 0.1, ¹N = 286; ²N = 240; ³N = 274 (original sample N = 404), Source: DEMES.

Throughout our analyses, we find no significant differences between men and women, refuting H2a and H2b. Of course, these results could indicate that minoritized migrant-background women’s greater numerical representation in parliament cancels out their lower engagement with communal organizations (Michon and Vermeulen 2013; Tillie 2004), resulting in null effects. Presently, however, we can only conclude that perceptions of the political representation of migrants are not gendered, which resonates to some extent with minor differences among the general population (Holmsberg 2020).

Concerning ethnic dividing lines, our results generally show that people with a Turkish background perceive the greatest political representation and people with a sub-Saharan African or Surinamese background least (see Figure 1). First, people with a Turkish background perceive the most political representation of migrant-background people (b = 0.40, b = 0.80, b = 0.83 and b = 0.81 in Table 2). The difference with the reference group (with a Surinamese background) is statistically significant and those with a Turkish background also perceive significantly better political representation than citizens with Sub-Saharan, Moroccan, East-Asian or Latin America backgrounds (all p < 0.05). The main effects hold for both types of representation (Model 3-4) and are robust to changing the measurements (Table A4, A5) or excluding denomination (A6). These results are in line with the privilege logic of Hypothesis 3a; Turkish Dutch people are numerically overrepresented in Dutch parliament, which seems reflected in their perceptions of migrant representation–even though they are regularly on the negative side of constructed boundaries.

Figure 1. Predicted perceived political representation by migration background.

Note: predicted perceived political representation based on Model 2 of Table 2, estimated for primary or secondary education, men, non-religious, first-generation and mean age. We distinguish four groups from top to bottom: (1) Turkey significantly differs (with at least p < 0.1) from Other MENA, Morocco, East-Asia, Other Latin America, Suriname and Sub-Sahara Africa; (2) Dutch Caribbean significantly differ from Suriname and Sub-Sahara Africa; (3) Other MENA, Morocco and East-Asia significantly differs from Turkey, Suriname and Sub-Sahara Africa; (4) Other Latin America significantly differs from Turkey; (5) Suriname and Sub-Sahara Africa significantly differ from all other migration backgrounds, except Other Latin America.

However, the same does not hold for people with a Moroccan background, although they are also numerically overrepresented in Dutch parliament. Their perceived political representation is better than people with a Surinamese and Sub-Saharan African background, but worse than people with a Turkish background (see Figure 1)⁸–a difference we did not expect in Hypothesis 3a. Although more research is needed to disentangle what is going on here, these results might signify that Moroccan Dutch people’s integration is more problematized than the Turkish Dutch’s. If so, this might mean that the privilege of numerical representation only leads to greater perceived representation among groups that are not exceedingly problematized in political debates.

Dutch citizens with an Other MENA or East-Asian background experience significantly more political representation than citizens with certain other backgrounds, such as Suriname or Sub-Saharan Africa. The difference of those with a Latin-American background is less straightforward. These citizens do perceive more substantive representation than those with a Surinamese or sub-Saharan African background (b = 0.32, Model 4 in Table 2), but they do not perceive significantly more descriptive representation. Overall, some of the group-dyadic results fit Hypothesis 3a, but of the calculated coefficients only around 25 per cent are in line with it. Thus, we reject Hypothesis 3b.

At the same time, it deserves note that those with a Latin-American and Dutch-Caribbean migration background relatively more often answer “do not know” (Model A5, Table A5), which resonates with our reasoning behind Hypothesis 3b. The radical right problematizes certain ethno-cultural groups more so than others. Although Latin American migrant-background citizens are considered “non-Western” in Dutch debates, this exclusionary discourse is far less clear. From this perspective, we do not consider it surprising that these groups more often do not know about the political representation as their political representation is less politicized and also given less attention if debate and media.

Although not hypothesized, we also find a relatively large positive effect of those with other MENA and Caribbean-Dutch background. With respect to those with a MENA background, we can propose that the political party DENK may contribute to such feelings of representation. The positive effect among those from Dutch-Caribbean people we find more difficult to situate.

With respect to religious dividing lines, we find that Muslims perceive less political representation than other (religious or non-religious) people, especially in terms of substantive representation. Most Muslims have a Turkish, Moroccan or MENA background, while these people perceive relatively high level of political representation of migrants. The effect of religion is therefore stronger controlling for ethnic background (see Model A7, Table A6 without controlling for ethnic background). These results support Hypothesis 4a. At the same time, while we find indications that more religious Muslims perceive even lower political representation (b = −0.61) than less religious Muslims (b = −0.55), these differences are not consistently statistically significant. So, although we find indications that those who identify more strongly with a religion they are excluded on the basis of, we do not find consistent support for Hypothesis 4b.

Finally, we find no consistent support for generational differences in perceived political representation, refuting Hypothesis 5. Although our bivariate models show that the descendants of migrants perceive less political representation than migrants, this effect is not present in the other multivariate models. Descendants of migrants are on average younger and more often theoretically educated than those who migrated themselves, which may explain why they perceive less political representation in line with the boundaries mechanism. After controlling for the other characteristics, children of migrants even perceive higher descriptive representation than migrants (b = 0.28, Model 3, Table 2), which echoes the knowledge and social and institutional learning mechanisms from the perceived representation literature (e.g. Holmsberg 2020, 414; Ramakrishnan 2005), suggesting that those more familiar with the democratic dynamics have more realistic expectations, leading better experienced representation. Or, more bleakly: those who expect less, accept more.

Discussion and conclusion

Although our insights into the causes and consequences of this numerical political representation are expanding, we know remarkably little about whether minoritized migrant-background citizens perceive that they and their interests are represented, while inequalities in such aspects of political belonging are considered vital to democracies (Holmsberg 2020). This study therefore addressed to what extent there are intersectional differences in perceived representation, by zooming in on differences between different socio-demographic groups of minoritized migrant-background people.

Concretely, this study illustrated that taking an intracategorical approach reveals important differences among migrant-background citizens, even in a setting of an election triggered by institutional racism and in the context of Covid policy responses, disproportionally impacting people with a migration background. Indeed, there are various dividing lines in perceived political representation that illustrate the diversity of migrant-background citizens, which remain veiled when comparing migrant populations to non-migrant populations as is often done in previous research. For example, we find that Muslim migrant-background citizens perceive less political representation than others. Acknowledging such differences reveals complexities and nuances needed to improve existing theoretical frameworks, particularly as we understand these differences in terms of ethno-religious boundaries and privileges (McCall 2005).

Our results show that both privileges and boundaries might shape minoritized migrant-background citizens’ perceptions of the political representation of migrants. Which is dominant depends on the studied intersection, although we find more indications for boundary mechanisms, as different groups who experience exclusions along existing boundaries are less positive about the representation of minorities. For instance, people with university degrees (who might be more aware of boundaries existing) and those who identify as Muslim (on the negative side of constructed religious boundaries) perceive less political representation of minorities. Additionally, groups whose ethnic backgrounds are less problematized and therefore less clearly excluded by boundaries as East-Asian, Latin-American and Dutch-Caribbean people seem less negatively affected in their perception of political representation, which is the more noteworthy as these election took place at the height of the covid pandemic, in which particularly East-Asian Dutch people endured more open racism (see Odag and Moskovits 2024; Walter and Glas 2024).

Altogether, this combination of results might suggest that the boundaries mechanism is at work most profoundly, particularly for the groups that are discursively excluded most in the public discourse also being the ones experiencing least political representation. Although more studies are needed to assess how and whether people also perceive boundaries, for instance through experienced exclusion or consumption of xenophobic media, the root of these boundaries in public discourse centers around Islam nowadays (De Lange and Mügge 2015), and we find clearly separate effects of migration-background and religion, the latter becoming more pronouncedly negative when controlled for migration background. This clearly aligns with the boundary logic. At the same time, the privilege mechanism helps to understand why we find a more positive experienced representation among Turkish-Dutch citizens (certainly when controlled for religion).

Regarding, the (relative political) privilege of the Turkish-Dutch the actual, and visible political representation might matter. Only since 2021 have the Netherlands had a Muslim MP wearing a scarf (Mügge, Runderkamp, and Spierings 2024) while Turkish-Dutch MPs have been strongly visible, were responsible for found the ethnic-minority interest party DENK, and are well organized with interlocking organizations (Tillie 2004), which can and spread news about these MPs (Dagevos et al. 2024) and provide social learning (cf. Holmsberg 2020). This might help to increase the experiences of representation, and thus aligns with the logic of visible privilege. Notably, indication of privilege and learning mechanism were less clear among the Moroccan-Dutch. This groups is actually problematized most,⁹ which might imply that boundary mechanisms can cushion the effect of privilege as Moroccan MPs are also visible. How religion and ethnicity interrelate could be studied in more detailed and our result suggest privilege and boundaries provide building blocks.

Regarding privilege more generally, we do not find that minoritized women (privileged in terms of numerical representation) or more theoretical (i.e. “higher”) educated people (socio-economic privilege) perceive greater representation of minorities, implying that privilege hardly explains who perceives greater representation by itself. The absent effect for gender makes sense, as the discursive exclusion of migrant men might be stronger, but the discourse revolves around women’s position, and sexist media representations in general might make women feel ill-represented too (Haraldsson and Wängnerud 2019; Runderkamp et al. 2022). Similarly, women have long been overrepresented among migrant-background politicians, but their visibility as such was not more considerable (and at a local level and thus in networks less evident) also because these women positioned themselves as migrant women specifically. As our results suggest, these discursive elements are crucial and for the intersection of migration and gender, this is far from a simple situation.

Beside the gender differences, also the results for the Dutch Antillean background people are noteworthy. They perceive both more descriptive and substantive political representation of migrant-background people, compared to many other groups distinguished. This is not what we expected theoretically or would expect based on other results. Here, discussion on the Dutch colonial past that have become more prominent in the last few years might play a role, but this needs further attention, for instance by incorporating media consumption. Looking into this case in future work might refine our frame of understanding.

Speaking more empirically, a central contribution of this study is the analysis of new data on the perceived political representation among minoritized migrant-background citizens, and thus adding a subjective dimension to the notion of political representation (see also De Mulder 2022; Holmsberg 2020). As such, our results contribute to previous studies on objective political representation and larger concepts of political engagement and inclusion, for instance by showing that a strong (over)representation at least at the national level does not per se mean minority people’s perception is unequivocally positive, as we found for Moroccan-Dutch people. One avenue for further research in this respect is to expand the study of subjective representation beyond the representation of “those with a migration background” and exploring the role of one’s self-identification (see also Dagevos et al. 2024; De Jong and Mügge 2024). For instance, theorized mechanisms may apply more strongly when asked about groups one feels part of.

Another interesting direction for future research could be delving into different dimensions of representation. Although previous research has concluded that so-called objective substantive and descriptive representation are theoretically distinct (Pitkin 1967), our results show that patterns are strikingly similar when studying perceptions of descriptive and substantive representation (see also De Jong and Mügge 2022). At the same time, people seem to find the question on substantive representation easier to answer, while the discussion of what that entails is rather lively in the literature (e.g. Celis and Childs 2008). Particularly striking is also that people with a low occupation status stand out in terms of far more perceived substantive representation than perceived descriptive representation compared to the other occupational-status groups. This finding might be partly due to the lowest status group being rather small, but it might also be linked to supply-side differences: particularly minoritized voters from lower socio-economic strata kept supporting mainstream leftwing parties, while the mid-strata and less likely left-wing voters moved to the ethnic-minority interest party DENK (see Lubbers, Otjes, and Spierings 2023); might particularly this group be more optimistic about their interests being met by existing MPs even though they do not recognize themselves in the MPs? Future work could delve into such differences more.

For now, the use of migrant-specific data source allowed us to show the diversity and intersectionality among minoritized people, and we even could not explore all salient intersections (e.g. sexual orientation, health). We conclude that perceptions of political representation are shaped negatively by exclusions grounded in discursive boundaries, while actual visible representation might counter this; assessing differences among minoritized migrant-background citizens offers a more comprehensive understanding of these matters of political representation.

Authors contribution

All authors contributed equally to this manuscript and are listed in alphabetical order.

Disclosure statement

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

Notes

1 This term indicates a focus on international migrants and their children, who are minoritized or ethnicized as being different from the supposedly “native” majority population. Practically, this roughly translates to citizens (whose parents were) born in Africa, Asia or Latin America. They are often assumed to be socialized in a markedly different culture, and they are often also visibly identifiable as “being different from the majority population”.

2 “Theoretical education”, or more commonly but problematically “higher education”, is a scale running from no education via only primary education to lower vocational and higher secondary to higher vocational and finally academic education. The scale ranges from basic or practical (elementary education) to theoretical (master degree) education.

3 Also, theoretically educated people have, on average, a better grasp on politics (Ramakrishnan 2005; Wüst 2011) and thus accept outcomes more, as long as existing rules and norms are adhered to. They might also be more aware of the proportional representation of some major ethnic-minority groups in the Dutch case. If of the “privilege hypotheses” only the education one is confirmed, this alternative explanation need extra consideration.

4 See Table A1; 19% had “don”t know” on both perceived descriptive (32%) and substantive (23%) representation:

5 “other” (n = 1) was excluded from the analyses.

6 VVD (conservative-liberal party), CDA (Christian-democratic party), D66 (liberal-democratic party), and ChristenUnie (Christian party).

7 Pattern remains stable in robustness analyses, e.g. including “don”t know” as “neither too little, nor too much” (Table A3).

8 Moroccan-background and Surinamese-background show no significance bivariate differences (Model 1) and no multivariate differences excluding denomination (Table A6). Moroccan-Dutch are more often Muslim, which is rather important for perceived representation,. Moreover, in the full multivariate model (Model 2), multicollinearity is not a problem: all denomination and ethnic-background VIF-values are below 3.

9 Two of the most infamous political claims on the topic are Oudkerks' (Labor) “kutmarokkanen” and Wilders” (PRR) “What do we want, more or fewer Moroccans?”