Space invaders and norm-politicians: how the media represent the intersectional identities of Members of Parliament

Abstract How do the media describe the intersectional identities of elected politicians? Our study focuses on parliamentarians in the Netherlands who fall outside the prevailing norm in politics: women and female and male ethnic minorities. Drawing on 2,783 newspaper articles published between 1994 and 2012 and matched samples, we find that the media structurally emphasize the identities of all parliamentarians who are not white men. Women politicians are more often described in terms of gender, ethnic minorities in terms of ethnicity and Muslim politicians in terms of religion. Ethnic majority men, meanwhile, are most often described by their political ideology. We find that this works already for one minority identity, as well as multiple identities. By continuously highlighting the identities of politicians that diverge from the norm, the media, we argue, paint pictures of women and ethnic minority politicians as different and out of place.


Introduction
Although ethnic minorities and women are making inroads into parliamentary politics in many Western democracies, the dominance of white men remains the norm (Inter-Parliamentary Union, 2022), reinforcing prevailing beliefs about who belongs in politics and who does not. As minoritized politicians must challenge deeply rooted patriarchal, racialized and heteronormative beliefs about what politics and politicians look like, they are often seen as 'space invaders' (Puwar, 2004) who do not quite belong. Their presence in elected office, however, reveals what has hitherto passed as the undisputed norm. In politics and political media coverage, this norm, as our study shows, continues to be the white male politician.
The media is a crucial place for the enactment of politics, how the public sees politics and politicians (Kahn, 1994;Strömbäck and Esser, 2014;Savigny and Yates, 2019), and where voters obtain political information (Van der Pas and Aaldering, 2020). In this article, we ask: how do the media describe the intersectional identities of elected politicians? We focus on Dutch politicians who defy the normwomen as well as male and female ethnic minoritiesand measure the intersectional effects of how they are described in the media. How much emphasis is there on their identity and deviance from the norm?
We use a unique data set to measure how Dutch parliamentarians have been described in five national newspapers over a period of 18 years (1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012). Drawing on a corpus of 250,000 newspaper articles paired with data from the 'Pathways to Power: The Political Representation of Citizens of Immigrant Origin in Seven European Democracies' study (Fernandes et al., 2016) we analyze 2,783 articles. We compare visible minority politicians with ethnic majority politicians who are otherwise similar. The matched sample consists of four groups: ethnic majority men, ethnic majority women, ethnic minority men, and ethnic minority women. By leveraging intergroup and intragroup comparison (McCall, 2005), we provide insight into the inclusion and exclusion of minoritized politicians in the media.
Drawing on the Netherlands, we find stark differences in media coverage of the identities of norm and minoritized politicians. Whereas the norm of whiteness and maleness goes unmentioned, the media structurally emphasize the space-invading identities of minoritized politicians: women are more often described in terms of their gender, ethnic minorities in terms of their ethnicity, and Muslim politicians in terms of their religion, while ethnic majority men are more often described by their political ideology. Surprisingly, this already holds for those with one minoritized identity, as well as those with two minoritized identities. Thus, we find that these categories apply separately, as well as intersectionally. By continuously highlighting the aspect of their identities that mark them as diverging from the norm, the media paint a picture of women and ethnic minority politicians as different and out of place.
We employ intersectional analysis (Crenshaw, 1991;Collins, 2015) to show how 'axes of identity such as race and gender are both mutually constituted and mutually constitutive' (Ward, 2016: 318). We understand intersectionality as both an empirical and theoretical project and contribute to the scholarship by studying the interplay of marginalized identity categories in media coverage (Gershon, 2012) that each matter in politically significant ways (Hancock, 2007). Theoretically, we build on scholarship that suggests that politicians with multiple minoritized identities can enjoy a 'complementarity bonus' (Celis et al., 2014) in political recruitment and thus in descriptive representation (Bjarnegård and Kenny, 2015;Erzeel and Celis, 2016;Gershon et al., 2019). Nevertheless, our focus on politicians already in office shows that identity-based inclusion and exclusion works differently once they have passed the gauntlet of recruitment.
Next, we formulate hypotheses on how the media use identity descriptions to describe men and women, ethnic majority, and minority politicians. We then present our methods, followed by our results, which reveal that media coverage of politicians' identities indeed works in exclusionary ways.
Norm or norm-defiant? Differences between and within groups of politicians Slowly but surely, historically excluded groups are entering parliamentary politics. As their 'ability to rule' is still questioned in many quarters, their presence is important (Mansbridge, 1999). The inclusion of previously excluded groups is furthermore a precondition for representative democracy to be truly representative (Phillips, 1995).
Intersectional theory posits that power is clustered around certain categories and exercised against others (Crenshaw, 1991(Crenshaw, : 1297 and that individuals with multiple minoritized identities suffer greater exclusion. Scholars have applied an intersectional lens to study how these hierarchies and inequalities work out in politics. For instance, how they affect political recruitment and representation or how identity categories differ between and within groups. Perhaps counter-intuitively, research suggests that candidates combining several marginalized identities (e.g., gender and ethnic minority) can benefit in terms of descriptive representation (Bejarano, 2013; as those who tick more than one box can enjoy advantages in recruitment over candidates only belonging to a single marginalized group. Ethnic minority women, for instance, can benefit from existing women's networks within political parties, which may lobby on their behalf for (higher) positions on party lists (Mügge, 2016). The result of these intersectional hierarchies is that ethnic minority women are better represented than their male counterparts, as suggested by evidence from the Netherlands, Spain, Sweden, New Zealand, and elsewhere Barker and Coffé, 2018; Mügge, van der Pas and van de Wardt, 2019). The process of recruiting candidates to run for elected office has been referred to as an 'intersectional puzzle' (Celis et al., 2014) where multiple identity categories can kill two birds with one stone.
But once elected, the 'complimentary bonus' is far from straight-forward. A study on USA Congress finds that the production of difference based on identities -'racing-gendering'entails 'silencing, excluding, marginalizing, segregating, discrediting, dismissing, discounting, insulting, stereotyping, and patronizing : : : used singly and in combination to fix women of color "in their place"' (Hawkesworth, 2003: 531). This echoes Puwar's (2004) portrayal of the 'space invader' as the 'outsider-within.' Once in office, the combination of several marginalized identities often works in exclusionary ways.
The media plays a crucial role in suggesting how politicians should behave and look like: '[a]s institutions, the mass media provide an organizational framework within which individuals interact in the context of set roles and established power hierarchies' (Downing and Husband, 2005: 9). The media is where people learn about and make sense of politics (Savigny and Yates, 2019). Journalists have the power to define situations; by uncritically employing categories such as 'race' or 'gender,' they participate in racializing and gendering as well as the confirmation and reproduction of identity descriptions (Downing and Husband, 2005: 5).
The media are not a 'mirror' that reflects what is in a neutral manner, they are rather a distorted one (Kahn, 1994). The media and its journalists make choices in all phases of the journalistic process (selection, presentation, and framing). Scholarship has convincingly shown that in all of these phases, race and gender are among the factors that influence those choices (Tolley, 2015: 22). This is referred to as gendered or racial mediation.
An intersectional lens offers new perspectives on media coverage. For instance, in the framing of the French extreme right woman leader Marine Le Pen, two frames seem incompatible. 'Soft' because of her gender, 'hard' due to her ideology and ideas (Snipes and Mudde, 2020). Given that minority politicians are not a homogeneous group, an intersectional approach reveals unexpected frames.
While few studies have analyzed both gender and ethnicity in media coverage, existing work suggests that ethnic minority women face more disadvantageous media treatment than ethnic majority women. The coverage of ethnic minority women politicians often focusses on their gender (Ward, 2016). The greater visibility of ethnic minority women in the 2010 UK general electionthey received almost twice as much attention as their ethnic majority counterpartswas a double-edged sword. Although their novelty indeed led to a 'visibility advantage,' media coverage tended to be more negative and narrowly focused on ethnicity and gender (Ward, 2017). The media focus on what is new and newsworthy (Harcup and O'Neill, 2017) and the entry of newcomers is indeed a novelty (Verge and Pastor, 2018). By contrast, the identities of white male politicians go unmentioned and unquestioned (Kahn and Goldenberg, 1991); they seem to have no ethnic or gender characteristics and 'figure simply as "individuals"' (Phillips, 2008: 448). Although the entry of space invaders renders visible what has so far passed as the uncontested norm, media descriptions of their identity place them in racialized and gendered hierarchies, even when these categories are contested (Yanow et al., 2016). Examining the workings of these categories within and between groupsas we do in this articlereveals the assumptions, hierarchies, and social meanings behind them. We now introduce our hypotheses on the mentioning of gender, ethnicity, religion, and ideology for normative and norm-transgressing politicians.

H1: Gender
Consistent with earlier studies on the gendered media coverage of politicians, we expect that the gender of women politiciansof both the ethnic majority and minoritywill be mentioned more often than the gender of their male counterparts. We further expect that distance from the norm on one axis (such as gender) will transfer to other axes (such as ethnicity), leading to cumulative marginalizing effects. While white men remain the norm, we expect ethnic majority women (one minority identity) to be deemed closer to the norm than ethnic minority women (two cumulative minority identities). We therefore expect that the media will highlight the gender of ethnic minority female politicians even more than for their ethnic majority counterparts. Based on these theoretical considerations, we derive our hypotheses on the media's use of gender identity descriptions for elected politicians: H1.1: Coverage of ethnic majority women will more often highlight their gender than for ethnic majority men. H1.2: Coverage of ethnic minority women will more often highlight their gender than for ethnic majority women.

H2: Ethnicity
The contestation over categories in the Netherlands has most prominently centered on the confusing, dubious, and racialized distinction made by the Central Bureau for Statistics (CBS) between people with 'Western' and 'non-Western' migration backgrounds (Yanow and Van Der Haar, 2013). National statistics have also relied heavily on the distinction between autochtoon and allochtoon, with the latter literally meaning 'emerging from another soil.' After extensive public debate, the term was replaced in 2016 by 'person with a migration background.' 1 While some of these markers are increasingly contested today, they are found in our sample, given its long timeframe. We use the distinction between 'Western' and 'non-Western' migration background to operationalize our study, precisely to render visible the marking as 'other' when we zoom in on politicians with visible, non-Western migration backgrounds.
Like their treatment of gender, we expect journalists to emphasize ethnic minority identities. We derive from intersectional theory the expectation that the media will even more vehemently 1 The Central Bureau for Statistics differentiates between first-generation (born abroad) and second-generation (at least one parent born abroad) migration background. emphasize the otherness of politicians who are minorities on a second axis, meaning that journalists will emphasize ethnicity more for ethnic minority women than for ethnic minority men.
H2.1: Coverage of ethnic minority men will more often highlight their ethnicity than for ethnic majority men.
H2.2: Coverage of ethnic minority women will more often highlight their ethnicity than for ethnic minority men.

H3: Religion
Since the 1990s, the political problematization of Dutch citizens with a migration backgroundand by extension their descriptive representatives, ethnic minority parliamentarianshas centered on religion in general and Islam in particular Vliegenthart, 2017, 2021). Rendered the norm, the 'Judeo-Christian tradition' has given shape to conservative nationalism in the Netherlands and Western Europe more broadly (Hemel, 2014), rendering 'Islam' and 'Muslims' as its most prominent 'other' (Vellenga, 2008). We thus expect politicians who are known to be Muslim to be more often described by their religious identity than for instance Christians, agnostics, or atheists. We expect religion-based coverage to extend to all ethnic minority politicians, i.e., beyond those who are publicly known to be Muslim. Migration and religion, in particular Islam, have been conflated by the public and journalists to such an extent that we expect all ethnic minority politicians (Muslim or not) to be more often tagged with their religion, marking them as outside the 'Judeo-Christian' norm (Blackwood et al., 2015;Hopkins et al., 2017;Amer, 2020).
H3.1: Coverage of politicians publicly known to be Muslim will more often highlight their religion than for non-Muslims.
H3.2: Coverage of non-Muslim ethnic minorities will more often highlight religion than for ethnic majority politicians.

H4: Ideology
It is crucial for parliamentarians that the electorate know what political ideology they stand for, making ideological identities arguably the most important the media might discuss. We understand ideology to be 'a set of values and beliefs that structures how people think their broader societies and political institutions should be organized' (Xydias, 2021: 58). But heightened media attention to the gender, ethnicity, and religious identities of minoritized politicians may come at the expense of more substantive, ideological identities. We take the prevalence of descriptions of ideological identity as a proxy for substantive (issue) coverage. Indeed, previous research has shown that women politicians receive less substantive issue coverage than their male counterparts (Aday and Devitt, 2001;Devitt, 2002;Dunaway et al., 2013;Goodyear-Grant, 2013;Fridkin and Kenney, 2014), thus limiting the information potential voters can base their choices on. This holds as well for ethnic minority politicians (Tolley, 2015). We expect a similar pattern and expect that the more nonideological identity descriptions a politician is saddled with, the less space there will be for their ideological identities. Building on our previous hypotheses, we expect that the ideological identities of ethnic minority and women politicians will receive less media attention and expect this attention deficit to be largest for female ethnic minority politicians: H4.1: Coverage of ethnic majority women will less often highlight political ideology than for ethnic majority men.
H4.2: Coverage of ethnic minority men will less often highlight political ideology than for ethnic majority men.
H4.3: Coverage of ethnic minority women will less often highlight political ideology than for (a) ethnic majority women, (b) ethnic majority men, and (c) ethnic minority men.

Methods and case selection
To study how the media describe the intersectional identities of minoritized politicians, we compare how the media cover parliamentarians with visible, non-Western migration backgrounds and politicians without such a background. We analyze the media representation of both groups by focusing on descriptions of identity such as motherhood or fatherhood, relationships, dual nationality, and (command of) the Dutch language. As our hypotheses illustrate, we expect certain descriptions that indicate politicians' social position or group membership to position them outside the norm, with multiple marginalized identities leading to greater distance from the norm. Some of these descriptors can change (e.g., relationship status) while others cannot (e.g., skin color).
The Netherlands is a least-likely case for the stereotypical media coverage of politicians given the relatively high number of elected politicians with 'non-western' migration backgroundsin several parliamentary terms, even relative to their share of the population. With large-scale postcolonial and labor immigration reaching back to the decades following the Second World War (Morales et al., 2017), the Netherlands has one of the longest and percentage-wise strongest representations of ethnic minority politicians in Europe (Fernandes et al., 2016). In comparison, the inclusion of women in politics has lagged behind, extending to the period of our study.
Our research period covers six parliamentary terms from May 1994 to September 2012. We start in 1994 when the number of ethnic minority politicians rose from one to six, thus becoming a more structural feature. The percentage of women in parliament was then around 30% . During the period under study, ethnic minority women were overrepresented, compared to their male colleagues, as they were benefitting from more favorable conditions in parties, such as well-organized women's networks (Mügge and Damstra, 2013). This would suggest a least likely case for strong attention to gender and ethnicity by the media, as diversity in parliament had become more normalized in numbers.
We use a data set with extensive information on individual parliamentarians collected in the project 'Pathways to Power: The Political Representation of Citizens of Immigrant Origin in Seven European Democracies' (Morales et al., 2017). To create the Pathways data set, researchers checked all parliamentarians to determine whether they were identifiable as a visible minority based on their name and photograph. Using this variable, we selected all individuals unambiguously coded as 'visible minority.' We did so as the literature suggests that the most consequential opposition is that drawn between 'Western' and 'non-Western' migration backgrounds, whereas migrants from, for instance, other European countries are less seen through the lenses of race and ethnicity (Rath, 1999). This yielded a sample of 46 ethnic minority politicians -20 men and 26 womeneach of whom we matched with two ethnic majority politicians. We chose this double sample design to enhance reliability. All matches include politicians of the same gender and party; if this resulted in multiple possible matches, the best match was selected based on overlap in active terms, closeness of starting time, similarity in relative average position on the party electoral list, and similarity in age and education (in order of descending importance). In addition, each ethnic majority politician could only be matched to one ethnic minority. If an ethnic majority politician turned out to be the best match for two ethnic minority MPs, he/she was replaced with the secondbest match. Nevertheless, four ethnic minority MPs were matched to a single ethnic majority MP as all MPs of the same party and gender were already included in the matched sample. In the end, the complete selection consists of 46 ethnic minority and 88 ethnic majority parliamentarians. Based on the Pathways data set, we use the religion of the MP, for which coders searched in newspaper articles and online (such as personal websites) to determine whether the politician's religion was publicly known. The main descriptive characteristics of the sample of MPs can be found in the online Appendix 3.
We draw on newspaper data for two reasons. First, the textual format and diachronic availability allow large-scale, systematic analysis. Second, newspapers as a medium reflect a relatively high level of editorial control over the final product, matching our focus on the portrayal of politicians by the media and its journalists. We included the five largest-in-circulation national daily newspapers -De Telegraaf, Algemeen Dagblad, Volkskrant, NRC Handelsblad, and Trouwwhich include both broadsheet and tabloid formats and span the ideological spectrum. We built our corpus by downloading all articles that mention a term related to parliamentary politics or a political party during the parliamentary terms under study. 2 This resulted in a corpus of 273,557 articles, 17,186 of which mention an MP. To maximize the utility of our coding resources, we excluded brief or superficial coverage unlikely to contain descriptions of identity. We did this by selecting articles that made at least three mentions of the name of one of our sampled MPs, including one mention of the full (first and last) name, which led to a set of 4,406 articles. For each MP, we coded up to 60 newspaper articles, taking a random sample in cases where there were more than 60 articles, leading to saturation of markers used. In total, this resulted in 2,783 articles to code.
To develop the codebook, we read a random sample of newspaper articles covering ethnic minority MPs, noting whenever an identity description was used and coding semi-deductively descriptions that emerged from the data, based also on previous research on Dutch ethnic minority MPs (Mügge and Schotel, 2017) and previous scholarship. We remained open to new codes emerging from the data. We then selected ethnic majority MPs and coded newspaper articles mentioning them using the previously defined codes and adding new ones when necessary until we achieved saturation. The codebook itself was constructed through a semi-deductive process and applied deductively to the whole set of newspaper articles, and where needed, elaborated, and expanded. This process allowed us to study a wide range of identity markers. Moreover, our coding was informed by a meta-analysis of gender differences in political media coverage (Van der Pas and Aaldering, 2020). While training four coders, we made some final adjustments to the codebook, which can be found in the online Appendix 1. We grouped the codes into six main categories: (1) ethnicity and nationality markers; (2) gender and relationship markers; (3) age; (4) social class; (5) religion; and (6) ideology. Intercoder reliability was established on a set of 100 articles coded right after training, and another set of 100 articles throughout the regular coding; the results can be found in Appendix 2. Eight codes (autochthonous, heterosexuality, fatherhood, old, atheist, secular, Jewish, other religions) were so rare that they did not occur in the reliability sample. As intercoder reliability is therefore unknown, these should be interpreted with caution. The frequency of codes can be found in Table 1.
We use OLS regression on the data set of 2,783 articles to predict the occurrence of each code, with clustered standard errors for each group of MPs. The main variables of interest are a dummy indicating whether the MP mentioned in the article has a migration background, a dummy for the politician's gender, and the interaction between these two terms. We present the predicted outcomes for the four groups in graphs in the results section; the full regression tables are in the 2 The search string used in LexisNexis to create the corpus was (Tweede Kamer! OR ((de kamer) AND CAPS (Kamer) supplementary materials. As the dependent variables are (overdispersed) counts, we estimate negative binomial models, also in the supplementary material (Appendix 5).
The matched sample allows us to compare the media's political coverage of ethnic minority men with ethnic majority men and ethnic minority women with ethnic majority women. Note that ethnic minority men are not necessarily comparable to ethnic minority women. To ensure that differences are not driven by other factors, we control for the electoral term (using term dummies), the starting year of the MP, electoral list position, education, age, party, and year of article publication.

Results
How are the intersectional identities of elected politicians described in the Dutch media? Figure 1 displays six broad identity categories. H1 predicted that the gender of ethnic majority women will be mentioned more often than for ethnic majority men (H1.1) and that the gender of female ethnic minority MPs will be mentioned most of all (H1.2). Figure 1 shows some initial results with the broad category 'gender and relationships' pointing to the frequency of gender identity, relationship, sexuality, and parenthood markers. The figure displays 83% confidence intervals to show the statistical significance of the differences between the groups (Goldstein and Healy, 1995). These results are in line with Hypothesis 1, although there is no statistical significance. Ethnic minority women are most frequently described in terms of their gender, sexuality, or family role (42.3 times per 100 articles), while ethnic majority women come second with 22.4 mentions per 100 articles. Ethnic minority men are least often described by gender (6.2 times per 100 articles), while the gender, sexuality, or family role of ethnic majority men is mentioned 16.6 times per 100 articles. Figure 2 specifies these results, allowing us to address Hypothesis 1 more directly by isolating gender-related mentions. Within the code 'man,' we find descriptions such as 'statesman,' 'sympathetic man,' and 'passionate man.' For women, we find 'a beautiful woman, utterly feminine' and 'female minister.' What jumps out is the difference between genders, with both ethnic majority and minority women being described significantly more often by their gender than ethnic Note: OLS regression estimates with clustered standard errors, controlling for parliamentary term, year the MP was first elected, average list position, age, education, year, and party. Full regression tables in Appendix 4. minority and majority men. This partially supports our hypothesis: ethnic majority women get more gender mentions than ethnic majority men (H1.1), but ethnic minority women do not get significantly more gender references than ethnic majority women (H1.2). Contrary to our intersectional expectations, we do not find firm evidence that deviance from the norm on one axis (gender) translates to the second axis (ethnicity). Nor do we see significant differences when it comes to relationship status, sexuality, or parenthood, with sexuality hardly being mentioned.

H2: Ethnicity
We expected journalists to use ethnic descriptions more often for ethnic minority men than for ethnic majority men (H2.1), and more often still when describing ethnic minority women (H2.2). Looking at both the broad results in Figure 1 and the split-out results in Figure 3 which show the different codes included in this category, we find clear support for the first part of the hypothesis (H2.1) but not for the second (H2.2). As Figure 1 shows, ethnic majority women receive 8.2 ethnicity/nationality markers per 100 articles, while ethnic minority women receive the most, at 55.8 per 100 articles. The ethnic origins of ethnic majority men are mentioned 7.8 times per 100 articles, and for ethnic minority men, 52.2 times. There is a strong effect along the single axis of ethnicity, but we find no intersectional effects when it comes to differences between genders. Figure 3 further unpacks our findings according to the use of specific markers. Strikingly, ethnic minority MPs are more often described with the code 'Dutch' than ethnic majority MPs. This includes mentions of having or having obtained Dutch nationality (Nederlanderschap) or a Dutch passport (haar Nederlandse paspoort). These mentions may indicate that being part of the norm (being considered ethnically Dutch) is not worth mentioning, while deviance from the norm (for instance being Turkish-Dutch) is highlighted through the usage of hyphenated identities (Mügge and Schotel, 2017). The code 'regional Dutch' is most often combined with 'non-Western'; for the full co-occurrence table see Appendix 3. The code 'non-Western,' most often combined with 'migrant,' stands out as the most often used identity description for both ethnic minority men and women: 23.5 times per 100 articles for men and 21.6 times per 100 articles for women. Within these codes, we mostly find descriptions of countries such as 'born in Turkey' or 'of Moroccan descent.' Other descriptions frequently applied to ethnic minority MPs are 'migrant,' 'allochthonous' and, to a lesser extent, 'refugee. ' We also find a significant difference in the use of the term 'dual citizenship,' which is applied significantly more often to ethnic minority women than to ethnic majority women. Similarly, 'refugee' appeared significantly more often to describe both female and male ethnic minority MPs than ethnic majority female MPs (but not significantly higher than for ethnic majority male MPs). The term 'Dutch language' appeared significantly more often in covering both ethnic minority male and female MPs than for ethnic majority male MPs (but not significantly more than for ethnic majority women). Finally, race was rarely mentioned in the Dutch newspapers under study, although the marker appeared 2.1 times per 100 articles about ethnic minority women, significantly more than for ethnic majority politicians.

H3: Religion
We expected politicians publicly known to be Muslim to be described more often in terms of their religion than their non-Muslim colleagues (H3.1) and non-Muslim ethnic minority politicians to be identified more often with religion than their ethnic majority counterparts (H3.2). While Figure 1 indicates that ethnic minority MPs are more often described by religious identity descriptions, this does not tell us whether this holds for Muslim and non-Muslim minorities alike. Figure 4 addresses the hypothesis directly by splitting out religious identity descriptions along ethnic and religious lines. This shows that the media rarely uses religious identity descriptions except to mark Muslim MPs, who receive at least five times as many religiosity markers than any other group. We thus find support for H3.1. But contrary to our expectation in H3.2, the journalistic interest in religiosity does not spill over to non-Muslim ethnic minority MPs, whose religion is discussed equally often as for ethnic majority MPs.

H4: Ideology
The results show that media coverage tends to highlight any deviance from the norm: women MPs are more often described by their gender, ethnic minority MPs more often by their ethnicity, and Muslim MPs more often by their religion. H4 predicts that ethnic majority men will be described most often by their political ideology and ethnic minority women the least, with ethnic majority women and ethnic minority men falling in between. The ideology codes include mostly short descriptions such as 'left-wing,' 'right-wing,' 'social-democrat,' and 'liberal' as well as more specific markers such as 'populist' and 'pacifist.' Figure 1 demonstrates that ethnic majority men receive the most ideology markers: 5.3 times in every 100 articles (83%CI 4.4 -6.25). This is significantly more than for ethnic majority and minority women, who, respectively, receive 3.1 (83%CI 2.2 -4.01) and 1.9 (83%CI 0.5 -3.3) ideology markers per 100 articles. Ethnic minority men in our sample were also less often described using ideological identities than ethnic majority men (2.7; 83%CI 0.8 -4.6), but the difference is not statistically significant. While ethnic minority women were indeed described least often by their ideology, the difference to ethnic minority men and ethnic majority women is not significant. While ethnic majority men do stand out for their ideology descriptions in the media, our data does not reliably allow us to say whether the differences among the other groups generalize beyond our sample. Even though 'age' and 'social class' are not included in our hypotheses, they are included in our open coding (see Figure 1). Social class is highlighted for minority MPs, such as former occupations. Regarding age there are no big differences between groups.

Conclusion
Women and ethnic minorities are slowly but surely entering representative bodies worldwide. Intersectional scholarship has sought to reveal when, how, and under what conditions processes of inclusion and exclusion work for and against space invaders. We asked: how does the media cover the intersectional identities of politicians who fall outside long-established norms?
For our period under study, our research shows that media coverage of elected politicians works to highlight deviance on all axes of identity. Both ethnic minority and majority women are described significantly more often in terms of their gender, ethnic minority women and men in terms of their ethnicity, and politicians known to be Muslim in terms of their religion. Male MPs from the ethnic majority seem to have few ethnic or gender characteristics, they still pass as the uncontested norm and receive most substantive coverage, seen through the ideology markers. The backgrounds of elected politicians who fall outside the white masculine norm are continuously emphasized, thus marking them as different and out of place.
Contrary to our expectations, we found limited intersectional effects, although ethnic minority women are hit twice, both for their gender and their ethnicity. We may speak of 'exceptional disadvantages' of a systematic nature for female ethnic minority politicians, as has been found in previous scholarship (Ward, 2017). However, our study shows that exclusion of space invaders works on each axis independently as well, rather than on cumulative intersectional identities. One minoritized identity is 'enough' to experience racialized or gendered mediation.
Emphasizing the identity of newcomers by the media is not new or surprising, but what happens when the identity keeps being emphasized, or even leads to stereotyping? Such a 'doubleedged sword' is worrisome where it takes away time and place for substantive coverage, on which voters make their decisions. Additionally, 'newcomers' of multiple marginalized identities are not so much new anymore but have become a structural feature of parliaments. Such patterns continue to exist today, likely impacting those MPs with one or more 'space invading' identities that have entered parliament since our period under study, such as the first transgender woman MP (Lisa van Ginneken) or the first hijab-wearing MP (Kauthar Bouchallikht).
Our research proposes new strategies to empirically study and operationalize intersectionality, which has proven difficult in practice (McCall, 2005;Weldon, 2008;Celis and Mügge, 2018;Mügge et al., 2018). Combining qualitative and quantitative methods to capture as many identity descriptions as exist in the real world, constructing a codebook through an open, inductive process, and our use of matched samples are important strategies to put intersectionality into research practice. While we focused on religion and ideologybeyond the more common considerations of race/ethnicity and genderfuture research can consider other categories to capture a more complete image of identities used by the media.
Our study has important implications: where some scholarship has focused on the influx of more diversity into politics, we shed light on the problematic day-to-day reality of those that were able to enter the political realm, which includes their media coverage. We find that norm-defying identities are continuously emphasized, also beyond intersectional patterns. The ultimate consequence of the media continuously marking space invaders is the unequal conditions under which these politicians are able to represent their constituents, be re-elected and advance politically.