Intersectional representation
The introduction of a new approach to analyzing congruence
between the policy preferences of voters and their representatives
Name: Saskia Kleinendorst Student number: 10571744 Supervisor: dr. A. (Afsoun) Afsahi Second Reader: dr. D.J. (Daphne) van der Pas Date: 21-06-2019 Master thesis political science Specialization: Political Theory
Table of contents
1. Introduction 3
2. Political representation and Intersectionality 7
3. Methods and Data 16
4. Results 21
5. Conclusions 30
3
1. Introduction
“When it comes to Brexit, politicians are only respecting the ‘will’ of the white people”- “Record number of black, Asian and ethnic minority MPs elected to parliament” – “Labour selects just six BME candidates across 99 target seats” (Andrews, 2019; Kentish, 2017; Courea, 2019). When headlines become paradoxical about the representation of ethnic minority groups, by claiming record numbers of representatives from an ethnic minority background without actually representing the ‘will’ of these groups, alarm bells should start to ring about the scientific approaches we use to analyze political representation.
However, the traditional approach to questions about the political representation of marginalized groups has mainly been concerned with the descriptive and/or substantive representation of those that are ought to be represented (Celis, Child & Curtis, 2016: 813). Within this scientific field, scholars in favor of descriptive representation analyze whether or not there is a sufficient amount of representatives (if any) that share descriptive characteristics with the marginalized; while those in favor of substantive representation analyze whether or not there are elected representatives that act to a certain extent in accordance to the wishes of ethnic minority groups. These kinds of traditional research on political representation have proven somewhat problematic, at both the theoretical and empirical level, especially when it comes to analyzing those who face multiple types of marginalization (ibid).
The following three, which will be discussed in more detail in the subsequent section, are the main problems with the traditional approaches. First, this kind of research oftentimes treats gender and race as two standalone groups, which causes the discourse to be around the experiences of white women and black males (Cho, Crenshaw & McCall, 2013: 791). Secondly, even in research that acknowledges the intersections of gender and race, most of the empirical data that is available does not specifically ask the respondents about their ethnic background. Seeing these aforementioned groups as merely a categorical label ignores the substantive
4
content of different ethnic minority groups (Simon, 2012: 1381). Finally, most studies on the political representation of marginalized groups measure the level of representation by comparing the position on the left-right scale of the government to the positions of its citizens, which ignores the distribution of the within-group multitude of policy preferences (Schakel & Hakhverdian, 2018: 446; Golder & Stramski, 2010: 95).
Thus, due to the inability of the predominant methods of analyzing the political representation of those who are being marginalized, this article makes two important interconnected contributions to the field of political science. The first contribution is a theoretical one, which, in order to tackle the problems and limitations of substantive/descriptive representation, argues for the introduction of a new concept: ‘intersectional representation’. This represents the idea that elected and non-elected representatives in democracies should represent the preferences, to the highest possible level of congruence, of all its citizens, including those with multiple dimensions of disempowerment. The second contribution is empirical, which argues that the traditional approaches to empirically analyzing political representation are unable to establish the actual level of underrepresentation regarding the policy preferences of ethnic minority women. This article, therefore, uses the recently introduced many-to-many approach to analyze the level of intersectional representation (Golder & Stramski, 2010). In short, the many-to-many approach analyzes the number of similarities between distributions of preferences of citizens with different combinations of gender and ethnic backgrounds, and representatives’ preference distributions on any given policy issue. By doing so, analyses become more inclusive, because the substantive content of different ethnic minority groups is acknowledged instead of pushed aside.
For this research, the United Kingdom is used as context. Before the 2010 general election, Dawn Butler and Diane Abbott had been the only non-white women seated in the British House of Commons (Browning, 2019: 9). After the election, the media reported record
5
numbers of representatives with a minority ethnic background in positions ranging from local council members to members of parliament (Kentish, 2017; Wilson, 2017). However, a closer look at those ‘record numbers’ shows a completely different story.
From a descriptive perspective, the number of black, Asian and mixed ethnic (BAME) representatives is indeed on the rise, but the steepness of that increase is not that significant as the media portrays it to be. From the 650 members of the House of Commons, the number of MPs with an ethnic minority background increased from 2.3 percent to 4.4, while the number of BAME citizens increased from 9 percent to almost percent according to the 2011 Census (Office for National Statistics, n.d.; Browning, 2019; Ward, 2017: 45). Additionally, when you add gender to the mix, the number of female BAME representatives shows yet another skewed ratio. The percentage of female representatives in the House of Commons from a black, Asian, or mixed ethnic background was only 1.7%, compared to 2.5% of male BAME representatives.
From a substantive perspective, the UK serves as a difficult case as well. For example, it is estimated that by 2051 almost 27 percent of the population will be from an ethnic minority background (Browning, 2019). In the 2016 referendum concerning the withdrawal of the United Kingdom from the European Union, 73% of Black and 67% of Asian citizens voted for the “Bremain” (Andrews, 2019). However, the media mainly showed the Scottish, Northern Irish and the Democratic side of the “Bremain” story, while the voices of ethnic minorities were being silenced in almost all of the media coverage and political debate (ibid). Considering the fact that the next generations of Britain will be increasingly diverse, the underrepresentation of the minority population on a decision that has unprecedented influence on the future seems unjustifiable.
Thus, as the world is increasingly becoming aware of the unconscious and conscious negative racial and gender biases, research on political representation needs to become more inclusive as well. Especially when the level of descriptive and substantive representation begin
6
to contradict each other. Without specifically looking at the intersectional representation of those who face multiple dimensions of disempowerment, research on political representation would risk becoming an outdated scientific field.
The empirical findings of this current study provide us with additional evidence for the need for the new concept of ‘intersectional representation’. The analyses reveal stark differences between the congruence of policy preferences of politicians and those of the voters, categorized by different combinations of gender and ethno-racial background, indicating that the representation of those with one or multiple dimensions of disempowerment, is significantly lower on the four analyzed policy issues. Furthermore, the analysis shows that both gender and ethno-racial background contribute to unequal representation overall, but, in the case of the United Kingdom, the negative bias is aimed at both men and women with a BAME background, instead of only against women.
Overall, the key question addressed is to what extent the policy preferences of ethnic minority women in the United Kingdom are congruent to the policy preferences of their representatives when compared to those with different intersections of race and gender. In order to answer this question, the first section deals with the theoretical underpinnings of intersectional representation, by discussing the problems and limitations of the traditional approaches to representation. This is followed by a section explaining the many-to-many congruence approach to intersectional representation and placing it within the context of the United Kingdom. The third section presents the statistical findings of the research, focusing explicitly on the representation of those previously ignored in and excluded from traditional frameworks. The last section draws upon the entire thesis, tying up the theoretical and empirical strands in order to emphasize the importance of the concept of intersectional representation, and includes a short discussion of the implications of the findings to future research.
7
2. Political representation and Intersectionality
Representative government has not always been democratic government but both concepts seem nowadays to be inextricably linked to each other (Manin, 1997: 196). In modern-day democracies we expect our representatives to listen and act upon our needs and wishes; we expect them to represent us. But the act that is carried out by our representatives could be interpreted in many ways. According to Hanna Pitkin, representation in its most basic form means ‘the making present in some sense of something which is nevertheless not present literally or in fact’ (Pitkin, 1967: 9). The making present of A by B is a commonly used formula, like Dahl’s from Weber, which means that “if a is absent, he is not present: he is thought, conceived, imputed to be present in B” (ibid).
The materialization or application of the aforementioned formula could pan out completely different depending on what is being made present. For some writers, representation is an act of authority or accountability, for some, it depends on the representative’s characteristics which he shares with those he represents (Pitkin, 1967: 61). The latter is what we call descriptive representation, “in which a person or a thing stands for others by being sufficiently like them” (idem: 80). The representative ‘stands for’ something or someone absent by some corresponding features. Due to the fact that the representative is like the represented in relevant ways, the representative can give valuable information about the represented. That way the absent can be made present (idem: 81).
For those in favor, descriptive representation is a matter of accurate resemblance to the represented, and a requirement for justifying governmental action. Yet perfect accuracy of resemblance is near impossible when it comes to political representation. Therefore, representation by resemblance is always troubled by the question which characteristics should be doing the resembling. Should a representative have the same religious background or should gender, social class, or ethnicity be the shared characteristic?
8
Despite the fact that descriptive scholars are aware of the question which characteristics should be doing the resembling, they oftentimes do not take the complex relationship between multiple and conflicting experiences of subordination and power by different social groups into account (McCall, 2005: 1780). In 1989 and 1991, Kimberlé Crenshaw published a pair of essays, introducing ‘intersectionality’, an analytical frame with the intention of ending the treatment of race and gender as mutually exclusive categories of experience and analysis (Cooper, 2015: 2). Within this intersectional framework, it is mostly believed that “it is frequently the consequence of the imposition of one burden that interacts with preexisting vulnerabilities to create yet another dimension of disempowerment” (Crenshaw, 1991: 1249). Excluding those with a double marginalization would be simply paradoxical to representation because correspondence to the represented must be accurate (Pitkin, 1967: 63).
Therefore, regardless of the fact that decisions often times must be made by majority rule, minorities have a right to participate in the deliberations, that is, the representation (idem: 64). Anne Phillips emphasizes this point in her book The Politics of Presence, by stating that “in order to achieve more fair and adequate representation of those interest that were not explicitly consulted or debated during election campaigns, (…) it is vital to achieve that additional element of representation which arises from the presence of previously excluded groups” (1995: 45). By being part of the deliberations, it becomes possible for groups who are engaged in similar social justice projects to identify points of connections that can further both their own and the endeavors of the connected groups (Hill Collins, 1990: 37). “This framing – conceiving of categories not as distinct but as always permeated by other categories, fluid and changing, always in the process of creating and being created by dynamics of power— emphasizes what intersectionality does rather than what intersectionality is” (Cho, Crenshaw & McCall, 2013: 795).
9
However, in research on the political representation of marginalized groups, gender and race are often still treated as two standalone groups, which causes the discourse to be around the experiences of white women and black males (Cho, Crenshaw & McCall, 2013: 791). The simple way of treating different types of marginalization as parallel to each other overlooks the within-category diversity and the dynamic interaction between political institutions and different citizens groups (Hancock, 2007: 71). As a result, the descriptive approach creates an academic environment in which scholars unintentionally turn a blind eye to the people that face “the consequence of the imposition of one burden that interacts with preexisting vulnerabilities to create yet another dimension of disempowerment” (Crenshaw, 1991: 1249).
Yet even if there is a way to create a near perfect resemblance to the represented, this does not mean that these representatives will necessarily be the best at the act of governing; it is in no sense a matter of agency. Descriptive representation shows us our assumption that people’s characteristics are a guide to the action they will take and therefore introduces the importance of resembling one’s constituents. It still does not explain the actual activities of making representation (idem: 111 & 89). Where the representative is likened to a descriptive representation, he or she merely represents by how his/her characteristics are regarded; talking about his/her role or duties would make no sense. “Standing for” is a passive way of representation, there is no activity, no “acting for”. In the realm of action, the resembling characteristics of the representative are relevant only insofar as they guide his actions (idem: 142) We can only begin to speak of representation “when the substitute’s actions are, in some way or for some reason, to be ascribed to another” (idem: 139).
However, a classic controversy appears when the question is raised about the representatives’ level of independence. This ‘mandate-independence controversy’ asks whether the representative should be bounded by mandates, derived from the wishes of his constituents, or whether he should be free to act independently, as he sees fit (Pitkin, 1967: 145). A mandate
10
theorist will say that the representative should not be led by his own purpose; he can only act upon the wishes of those who sent him. Conversely, independence theorists will argue that a constituency is not one single unit, therefore a representative cannot simply act upon one wish, when there is not just one act to be carried out (idem: 147). Nonetheless, one can see the logic of both: it is not a matter of representing when the representative purposefully keeps doing the opposite of what his constituents would do, but at the same time, representing stops if the representative himself does nothing and his constituents act directly (idem: 151).
Substantive representation tries to battle this controversy by saying that “the representative’s obligation is to the constituent’s interest, but the constituent’s wishes are relevant to that interest. Consequently, the representative also has an obligation to be responsive to those wishes. He need not always obey them, but he must consider them, particularly when they conflict with what he sees as the constituent’s interest, because a reason for the discrepancy must be found. (…) The representative must act in such a way that, although he is independent, and his constituents are capable of action and judgment, no conflict arises between them. He must act in their interest, and this means that he must not normally come into conflict with their wishes.” (idem: 162 & 166). From the substantive view, political representation means that the representative must act in the interest of the represented, guided by his own judgment (Manin, 1997: 203). Being at odds with the wishes of the represented should not be the rule, but the exception.
Yet if the situation occurs where there is a conflict between the actions of the representative and the wishes of the represented, then the representative should always be able to give a good, rational explanation for his actions (Pitkin, 1967: 209). Here, the passive, descriptive ‘standing for’ is traded in for a substantive ‘acting for’. This does not mean that political representation is any single action by any one participant; political representation involves many people and groups. Thus, the relationship between the representatives and the
11
represented is not just a two-way street; the relationship is intermediated by political institutions, economic interests, civil society organizations and so forth.
Meanwhile, the individual legislator is a member of a complete representative body as well (idem: 224). For political representation it is required to have institutions, such as a representative body, that can secure a government responsive to the public interest and opinion (idem: 234). Thus, political representation is “a matter of substantive acting for [many] others [by many others], it requires independent action in the interest of the governed, in a manner at least potentially responsive to them, yet not normally in conflict with their wishes” (idem: 222).
Nonetheless, most studies on the political representation of marginalized groups measure the level of representation by comparing the position on the left-right scale of the government to the position of its citizens, instead of comparing the level of congruency of policy preferences between citizens and the collective body of representatives (Schakel & Hakhverdian, 2018: 446; Golder & Stramski, 2010: 95). By doing so, the position of both representatives and citizens is reduced to a mean, which ignores the distribution of the within-group multitude of policy preferences. This creates the problem that the representation of marginalized groups in a highly polarized electorate cannot be measured because simply looking at the mean position of the government versus its citizens seems to presume a centrist electorate (ibid). It does not analyze whether or not the governmental action is in the interest of the marginalized, or “in a manner at least potentially responsive to them” (Pitkin, 1967: 222). Therefore, ironically enough, this way of analyzing the political representation of marginalized groups seems to marginalize those who are already marginalized even more.
However, according to Jane Mansbridge, the substantive acting for the represented by the representatives would reach the highest possible level of congruence through both deliberation and aggregation (1999: 630). She argues that deliberation between representatives of different groups would lead to uniformity/consensus on policies that is congruent with the
12
preferences of their constituencies, while the aggregative function would secure that, in time of conflict, the interests of “every group whose interests conflict with those of others, in proportion to the numbers of that group in the population” are being represented (idem: 634). Yet Crenshaw argues that the experiences of women are “often shaped by other dimensions of their identities, such as race and class” and can thus not be classified as one, single category (Crenshaw, 1991: 1242). As a result, Mansbridge’s aggregative function would need a near perfect semblance to all of the represented in order to successfully represent all of the possible combinations of different dimensions of their identities.
Another point made by Mansbridge is that, instead of choosing one over the other, descriptive representation would further the substantive representation of interest by improving the quality of deliberation (idem: 654). In later work, she introduces not just one form of representation to battle Pitkin’s influential work, but she comes up with four different types of representation. These are not mutually exclusive per se, they may interact over time with each other. The four types do have one thing in common: the criterion of constituent-representative congruence (Mansbridge, 2003: 526). Rehfeld (2009) has challenged some of Mansbridge’s conceptual claims, arguing that the four types unnecessarily complicate empirical and normative research because she seems to simply trade the ‘mandate-independence controversy’ for her own controversy.
Despite their differences, both of them agree on the fact that the ‘mandate/ independence’ or ‘trustee/delegate’ debate has been too dominant regarding the proper relationship or congruence between representatives and their constituents. Consequently, the representatives’ source of judgment, aims of legislation and his/her own responsiveness to sanction have been neglected in research on representation (Rehfeld, 2011: 2). Secondly, Rehfeld and Mansbridge put forward the view that representation should be investigated systemically “in the sense of involving many different parts interacting with one another in
13
interesting and complex ways” (Rehfeld, 2011: 11). With their systemic view, they seem to combine descriptive and substantial representation by arguing for investigating the entire representative system as well as the resembling characteristics of elected and non-elected representatives (Mansbridge, 2011: 629). Just as Pitkin, they advocate for looking beyond the individual one-to-one relationship between the representative and the represented (Mansbridge, 2011: 621; Pitkin, 1967: 147; Rehfeld, 2011: 11).
However, most empirical research on representation does take the delegate/one-to-one side of the controversy. This kind of dyadic representation looks at the relationship between the individual representative and his or her constituents; how well the legislator can act as a delegate for his constituency on legislative decisions (Ansolabehere & Jones 2011: 1). The reason behind this favoritism is that dyadic representation is easier to measure: just compare for example roll call votes with public opinion surveys, where “good” representation means that the representative has a voting record that corresponds to his or her constituents’ preferences (Rehfeld, 2009: 219). Yet again, the position of both representatives and citizens is reduced to a mean, which ignores the distribution of the within-group multitude of policy preferences.
Additionally, Kymlicka (as cited in Urbinati & Warren, 2008) argued that the representation of individuals by individuals would not be sufficient, as self-identity depends on groups relationships and resources. This one-to-one approach ignores the fact that many groups within established democracies lack passive as well as active representation. “We need to understand representation as a relationship, mediated by group histories and experiences, through which relevant constituencies—particularly those related to fairness—come into existence” (Urbinati & Warren, 2008: 392). Oftentimes certain groups will get oppressed or marginalized, not because of a malicious dictator, but as a result of unconscious assumptions, bureaucratic mechanisms and political systems. We cannot get rid of these disadvantages and
14
injustices if we only look at an individual level and thereby dismiss the claims of the groups to be represented (Young, 2013: 277).
Furthermore, this dyadic one-to-one view could lead to a misrepresentation because it may lead constituents to base their vote on descriptive characteristics, while overlooking the ways in which they are being substantively misrepresented. “By emphasizing the descriptive aspect of the dyadic relationship at the expense of the substantive, policy-oriented aspect, constituents may misperceive the extent to which their interests are being represented and fail to hold their members accountable” (Ansolabehere & Jones 2011: 18). However, researchers need to be careful with their ‘unitary’ approach to marginalized groups, because then the focus is solely on one category or characteristic as the best explanation in particular situations of social disadvantage (Kantola & Nousiainen, 2009: 461). This approach could end up ignoring the inequalities of other mutually affected groups; causing marginalization of some and privilege for others (Lombardo & Verloo, 2009: 1).
Therefore, when looking at different groups, researchers should be aware of the presence of multiple intermediaries (Pitkin, 1967: 223). Merely treating different types of marginalization as parallel to each other (such as race and gender), without looking at the within-category diversity and the dynamic interaction between institutions and groups (Hancock, 2007: 71) produces an additive model. This so-called ‘multiple approach’ would lead to competition between the marginal groups, instead of a model that can genuinely change the logic of the system (idem: 70). Intersectionality remedies this problem by specifically focusing on these different groups and their own experiences by “shedding light on hierarchies and power relations both between and within groups” (Mügge & Erzeel, 2016: 504).
However, even in research that acknowledges, for example, the intersections of gender and race, most of the empirical data that is available does not specifically ask the respondents about their ethnic background. Most of the questions in surveys from the Global North simply
15
distinguish between non-immigrant, Western immigrant, and non-Western immigrant groups, even though these are spread out over a multitude of ethnicities. Consequently, even when an empirical analysis acknowledges different intersections, seeing these aforementioned groups as a merely a static, categorical label, instead of following the respondents’ self-identification, ignores the substantive content of different ethnic minority groups (Simon, 2012: 1381).
Thus, even though in recent years there has been an increasing amount of concern that intersectionality has outlived its analytic usefulness (Cooper, 2015: 2) and should not extend beyond research on groups outside of women of color (Alexander-Floyd, 2012: 19), it remains widely believed that without an intersectional framework, political research would become some sort of blunt uniformity instead of “a ‘universal’ that is contingent, provisional, and rich with particulars, that entails the dialogue of all, the autonomy of each, and the dictatorship of none” (Chun, Lipsitz & Shin, 2013: 924). Therefore, in order to incorporate the substantive content of different ethnic minority groups into politics, ‘intersectional representation’, the idea that elected and non-elected representatives in democracies should represent the preferences of all its citizens, is needed.
In addition, it is important to set this concept apart from Ange-Marie Hancock’s interpretation of intersectional representation. In a multitude of discussions and analyses, Hancock recommends conceiving intersectionality as “a body of research – a paradigm – rather than a content specialization” (Hancock, 2007a: 253). She argues that by using intersectionality as a research paradigm, it could potentially generate problem-driven research, which could move beyond earlier approaches to studying the problem (Hancock, 2007b: 75). Hancock suggests, based on prior work on intersectionality as a research paradigm, that researchers should incorporate the following moves into their empirical works: “(1) configure cases in a way that fully acknowledge the intersecting roles of racism, sexism, homophobia, and other processes of marginalization; (2) improve the operationalization of such processes by using
16
case-oriented rather than variable oriented empirical analysis; (3) incorporate the reality that there are multiple paths, even within race or gender groups, to the same outcome of interest; and, (4) where applicable, better incorporate the richness of narratives and other interpretive data into scalable policy proposals for social change” (Hancock, 2013: 285).
However, if you approach it that way, the effect of looking at intersectionality only as a research paradigm, is that it becomes necessary, in evaluating the level of representation, to use the concept of substantive representation or descriptive representation. This hinders researchers in understanding the political representation of structurally underrepresented groups, due to the fact that these concepts are, as mentioned before, not fit to look at multiple intersections of dimensions. These traditional approaches are only useful when considering gender and ethno-racial background as parallel to each other. Conversely, ‘intersectional representation’ is the idea that elected and non-elected representatives in democracies should represent the preferences, to the highest possible level of congruence, of all its citizens, including those with multiple dimensions of disempowerment. Thus, if one understands and supports the necessity of intersectionality as a research paradigm, he or she needs to begin by incorporating the concept of ‘intersectional representation’.
3. Methods and Data Many-to-what?
As previously noted, the second contribution is empirical, which argues that the traditional approaches to empirically analyzing political representation are unable to establish the actual level of underrepresentation regarding the policy preferences of people who face multiple dimensions of disempowerment. The reason behind this is that most empirical research does not engage with comparing the level of congruency of policy preferences between citizens and
17
the collective body of representatives (Golder & Stramski, 2010: 95). Despite the fact that looking at this type of a many-to-one relation between citizens and their government is highly relevant for political science, this approach has two major shortcomings when it comes to political representation. Firstly, the many-to-one approach is favorized by scholars because they believe that governments are the most influential when it comes to the policymaking process. The emphasis on governments is problematic for research on political representation since it ignores how accurately the collective body of representatives corresponds to the preferences of the citizens; it focusses on one form of representation and thereby ignores how substantively representative the legislature as a whole is (ibid).
Secondly, most studies on substantive representation measure the level of congruence between the representatives and citizens by simply comparing the stance on the left-right scale of the government to the position of its citizens. However, the interpretation of being ‘left-wing’ or ‘right-wing’ could differ significantly from each other, which makes it an uncertain measure. This problem is remedied by the many-to-many approach by evaluating how similar the distributions of citizen and representative preferences are on any given policy issue, which makes it impossible to test whether there are statistically significant levels of congruence between subgroups.
Thus, while within the field of democratic theory, the emphasis on the importance of having a collective body of representatives whose preferences are congruent with its citizens already has a long history, empirical research is still falling behind (Schakel & Hakhverdian, 2018: 446; Golder & Stramski, 2010: 95). Therefore, this research uses Golder and Stramski’s many-to-many approach, which means the analyses is focused on the amount of overlap between positions of many (representatives) versus the positions of many others (constituents). This is done by comparing the percentage of representatives that have positioned their own preference on score X on policy issue Y to the percentage of voters that have positioned
18
themselves on that same score. To get to the congruence score on that issue, the lowest percentages of the rest of the distribution are all added up. This results in a number that could range from 0 to 100 percent: “the higher the percentage, the larger the overlap. The larger the overlap, the better the views of the electorate are mirrored in parliament” (Schakel & Hakhverdian, 2018: 446).
Case selection
The United Kingdom was selected as the case for this paper. Their Westminster political system uses the first-past-the-post (FPTP) single-member plurality method. It is traditionally a two-party system, but the dominance of the Conservative and Labour parties has been challenged by the rise of the Liberal Democrats (Hague & Harrop, 2013: 286). However, what makes the United Kingdom interesting as a case, is the expectation that this kind of majoritarian systems will not produce policies that correspond to the diversity of citizen preferences because of its low number of political parties and parties’ tendencies to centrist positions (Golder & Stramski, 2010: 100). Therefore, low levels of congruence between citizens and their representatives are to be expected within the British context (idem: 104). In a sense, this makes the UK-case a ‘most-likely’ case, which means that the UK is almost certain to fit the theory if the theory is true for any case at all (Levy, 2008: 12). Meaning that if the data does not show us lower levels of congruence for ethnic minority women in the United Kingdom, the theory about the negative bias against them would be damaged.
Data
To answer the question whether or not political representation of policy preferences in the United Kingdom is biased against black, Asian, and mixed ethnic minority (BAME) women,
19
this research needed three different datasets: the British Election Study (BES), the Ethnic Minority British Election Study (EMBES), and the Comparative Candidate Study (CCS), all three conducted in 2010.
While the BES is one of the longest running social surveys in Britain and gives us robust data of the British citizens at the national level after each general election, conducting a thorough research on the representation of BAME women would not be possible without a boost of ethnic minority respondents (Fisher, Heath, Sanders & Sobolewska, 2010). The absence of ethnic minority respondents is a common problem in comparative research, because most population statistics collect no information beyond birthplace (Simon, 2012: 1376). Only using the respondents’ place of birth would be an imperfect measurement to identify the respondents’ ethno-racial background because firstly, some might not agree with their assignment to a particular group and secondly, some might be foreign-born but at the same time might have ‘national’ parents (Simon, 2012: 1375-1381; Bloemraad, 2013: 654).
The BES and EMBES tackled this problem by asking the respondents to which group they considered themselves to belong (Sanders & Whiteley, 2014; Fisher, Heath, Sanders & Sobolewska, 2010). In both surveys, the respondents were given fifteen different options and an additional write-in box, for those who considered themselves having a different background than the given options. By wording it as ‘belonging to’, this approach acknowledges the substantive content of minority groups, instead of using minorities just a categorical label (Simon, 2012: 1381). The EMBES mainly focused on including respondents from the five biggest ethnic minority groups. For the UK, these were Black Caribbean, Black African, Indian, Pakistani and Bangladeshi (Fisher, Heath, Sanders & Sobolewska, 2010: 4). By merging the BES and EMBES datasets together, each (minority) group consisted of a sufficient amount of respondents, so that claims could be made.
20
The third dataset is the UK General Election Candidate Survey from the Comparative Candidate Study. The CCS was created as an international effort to create a cross-nationally comparable dataset. For this research, the only data needed is that from the candidates who stood in the 2010 elections to the House of Commons in Great Britain. In total, 1472 of all 4042 candidates completed the survey, which is a response rate of 36.4% (FORS, 2016).
Variables
In order to see how congruent the policy preferences of the political representatives are with ethnic minority women, when compared to men and other ethnicities, the respondents of the (EM)BES needed to be categorized. The (EM)BES categorized the fifteen answer options of the question about their ethnicity into five groups: ‘white British’, ‘mixed ethnic’, ‘Asian’, ‘black’, and ‘any other ethnicity’. Due to the number of respondents in the ‘mixed ethnic’ and ‘any other ethnicity’ category these groups are combined. That way each group contains roughly the same amount of respondents. Secondly, for the intersectional congruence analyses it is important to compute the variables ‘gender’ and ‘ethnic (minority) groups’ together into one variable. By doing so, eight different categories are created so that congruence for all possible combinations could be calculated (see Appendix A).
As a result of using datasets from different institutions, the questions asked are not entirely consistent. However, the same questions were asked in all three surveys on four policy issues. These were on the topics of integration, tax expenditure, military interventions, and immigrants’ influence on the economy, all but one measured on a five-point scale (see Appendix B for the survey questions). The cultural/social and economic policy dimensions are being satisfied by these question; the question on immigrants’ influence on the economy even fits both dimensions.
21
4. Results Descriptive statistics
Table 1 illustrates the general descriptive statistics for four policy issues for both the preferences of the citizens and the preferences of their representatives. It is apparent from this table that the average preferences of the citizens differ from those of the politicians. The most significant difference between both groups is in response to the question of whether immigrants were generally good for the country’s economy. Citizens tend to disagree, while representatives believe that immigrants have a positive influence on the economy. The least significant difference can be found between the policy preferences on the issue of ‘tax expenditure’. The average citizen, as well as the average representative, prefers policies that would provide a stable network of social security.
TABLE 1. General descriptive statistics for four policy issues
Mean Std. Dev. min max N
Citizens
Integration 2,62 1,05 1 5 5685
Tax expenditure 2,65 0,95 1 5 5479
Immigrants and economy 3,22 1,11 1 5 5579
Military intervention 3,74 1,17 1 5 5413
Representatives
Integration 2,35 1,18 1 5 1431
Tax expenditure 2,82 1,12 1 5 1432
Immigrants and economy 2,51 1,23 1 5 1421
Military intervention 3,09 1,28 1 5 1422
A second statistical analysis was used to compare the policy preferences of different groups within British society, without looking at the preferences of the representatives (see Appendix A for the categorization). From the data in Table 2a and 2b, you can see that the differences between the policy preferences of different ethnic groups and genders are clearly distinct, but
22
some differences between preferences seem to be negligible (for example women on the topic of military intervention [Figure 2b]). The most significant difference when it comes to policy preferences is yet again on the topic of the ‘influence of immigrants on the economy of the United Kingdom’. This is most noticeable when the average position of ‘mixed ethnic men’ is compared to the average position of ‘mixed ethnic women’ (difference of 0,35).
TABLE 2a. Average positions on policy issues of men
White Asian Black Mixed Ethnic
Integration 3,05 2,15 2,34 2,58
Tax expenditure 2,59 2,88 2,74 2,77
Immigrants and economy 2,89 3,60 3,90 3,56
Military intervention 3,57 3,77 3,45 3,55
TABLE 2b. Average positions on policy issues of women
White Asian Black Mixed Ethnic
Integration 2,97 2,09 2,32 2,69
Tax expenditure 2,41 2,89 2,70 2,54
Immigrants and economy 2,71 3,43 3,77 3,21
Military intervention 3,89 3,90 3,74 3,80
In order to see how significant differences in positions taken by the groups with different intersections of race and gender on these four policy issues are, a multiple regression analysis was done, for which the ‘white males’ were being held constant (as shown in table 3). As one might expect, there were significant differences found between ‘white males’ and all the other combinations of race and gender on the two topics of ‘integration’ and ‘the influence of immigrants on the economy’ However, no significant differences in policy preferences were found on the topic of ‘tax expenditure’ between men from a mixed ethnic background and women from black or mixed ethnic background. Additionally, no significant differences were found on the policy issue of ‘military intervention’ between ‘white males’ versus men and
23
women belonging to a mixed background. This means that these groups roughly share the same policy preference, when compared to the policy preferences of ‘white males’.
TABLE 3. Level of significance on policy issues, for which ‘white males’ were being held constant
Integration Tax expenditure Immigrants and economy Military intervention
Sig. Sig. Sig. Sig.
Male Asian 0,000 0,000 0,000 0,001 Male Black 0,000 0,047 0,000 0,018 Male Mixed 0,000 0,291* 0,000 0,756* Female White 0,079 0,000 0,000 0,000 Female Asian 0,000 0,000 0,000 0,000 Female Black 0,000 0,141* 0,000 0,019 Female Mixed 0,001 0,555* 0,011 0,111*
*: For these combinations, no significant differences were found when compared to ‘white males’
Additionally, to prevent ignoring the distribution of the within-group multitude of policy preferences, by using the categories as shown in the table above, an additional analysis was done to check for significant differences within these groups. This has to be done because for the empirical side to ‘intersectional representation’, it is required to check whether or not the differences within these groups are too big/small enough to consider analyzing them as a categorical unit. It should be mentioned beforehand that the presence of significant differences does not mean that is it no longer possible to draw conclusions based on these groups. On the contrary, its presence only emphasizes the need for ‘intersectional representation’. If you want to compare different combinations of disempowerment, you should not only be aware of the within-group multitude of policy preferences, but you need to check for these differences as well.
First, the ‘Asian’ category consisted of the following answer options: Indian, Pakistani, Bangladeshi, Chinese or any other Asian background. This means ten different combinations had to be checked in isolation for possible significant differences on all four policy issues,
24
therefore, resulting in 40 different checks. Interestingly enough, the analysis revealed that only on the issue of ‘military intervention’ do the positions differ from each other. These significant differences were found between four out of ten combinations: Indian versus Pakistani, Indian versus Bangladeshi, Pakistani versus Chinese, and Bangladeshi versus Chinese.
Secondly, those who felt that they belonged to a ‘black’ background could select between black Caribbean, black African, and any other black background. In its turn, these categories created three different combinations. The only significant differences were found when those who belong to the black Caribbean category were compared to the black African category on the issues of integration and the influence of immigrants on the economy. This should be interpreted as that those who feel like they belong to the black African category, in contrast to the black Caribbean category, tend to disagree more, when asked if they believe that immigrants have a good influence on the economy, and believe more strongly that immigrants should be required to adjust to the customs of the United Kingdom.
Lastly, the ‘mixed’ category consisted of the following options: white plus black Caribbean, white plus black African, white plus Asian, white plus any other mixed background, and any other mixed background. This means that another 40 checks had to be done in isolation; ten different combinations of ethno-racial groups on all four policy issues. Like the ‘Asian’ category, most of the significant differences were found on the issue of ‘military intervention’. An additional significant difference was found on the topic of ‘immigrants’ influence on the economy’ between ‘white plus black Caribbean’ and ‘white plus black African’. This analysis showed that those who feel like they belong to the ‘white plus black African’ category, are more inclined to disagree when asked if immigrants have a good influence on the economy.
25
Many-to-many
Prior to the intersectional many-to-many congruence analysis, two extra analyses had to be done: representatives versus ethnic (minority) groups, and representatives versus gender groupings. To determine whether the policy preferences of the different ethno-racial groups are congruent with those of their representatives, the percentage of representatives that have positioned their own preference on score X on policy issue Y was compared to the percentage of voters that have positioned themselves on that same score. The congruence scores are then calculated by adding the lowest frequencies for each position. Congruence between citizens and their representatives is high when the distributions of citizens and representatives’ preferences are similar; it is perfect (100%) when the two distributions are identical, and vice versa (Golder & Stramski, 2010: 104).
For example, on the left side of figure 1a, which indicates the believe that immigrants have a positive influence on the economy, the line for those with a white background lies below the line for representatives. As mentioned above, the lowest frequency for each position are summed; meaning that in this case, we use the percentage of the citizens from a ‘white background’ who places themselves on the first position (9.50%). For the second position, the line of those with a white background is still below the one of the representatives, so their frequency (33.50%) is added to the first number. After doing this for the rest of the distribution, the congruence level adds up to 71.3%.
A closer look at the different shapes of the lines and the amount of shaded area in figure 1a and figure 1b, reveals that there is a significant difference between the level of congruence of respondents with a white background and respondents with a mixed ethnic background (see Appendix C for a more detailed overview, including the numbers regarding the congruence levels of the other ethno-racial categories). The total congruence scores support this claim, because for those with a white background their score on this topic is 71.30% and the
26
congruence score for those with a black ethnic background is significantly lower: 48,4%. Put differently, the representation of black ethnic citizens is significantly lower on the topic of ‘immigrants’ influence on the economy’ than the representation of those with a white ethnic background.
FIGURE 1a. Congruence ‘Immigrants and Economy’ – Respondents with a white background
FIGURE 1b. Congruence ‘Immigrants and Economy’ – Respondents with a black background
However, if the data is analyzed based on gender, instead of looking at the differences between the congruence scores of ethnic (minority) groups, it becomes clear that there are no highly
27
significant differences between the congruence scores of men and women on these four policy issues (see Appendix D for a more detailed overview). The only policy issue worth mentioning in this analysis is ‘military intervention’, as shown in figure 2a and 2b. On the position of ‘neither agree nor disagree’ both men and women are rather close to their representatives, but from that point on the frequencies start to diverge again. In other words, male and female citizens agree to a certain point with their representatives but both have a strong policy preference that is against the involvement of the military against the Taliban.
FIGURE 2a. Congruence ‘Military intervention’ – Male respondents
28
Intersectional many-to-many
The final analysis, needed to answer the question of whether or not political representation of policy preferences in the United Kingdom is biased against black, Asian, and mixed ethnic minority (BAME) women, uses the ‘intersectional’ many-to-many approach. The frequency of each answer option is once again calculated and added up to each other, to establish the congruence level for all eight societal groups (see Appendix E).
A comparison between the levels of congruency on the four different topics reveals that British citizens are best represented when it comes to the policy issue of ‘integration’. The lowest score of congruence, and therefore the issue where citizens are the least well-represented, is found on the issue of ‘immigrants’ influence on the economy’, as illustrated in Figure 3. It shows that congruence increases at both ends of the ‘ethnicity’ axis (which corresponds to white and mixed ethnic), as well as at both ends of the ‘gender’ axis. This provides us with support for the claim that white and mixed white citizens are better represented than those with an Asian or Black background. However, there is not enough support for the claim that women are less well-represented than other groups, like those with an ethnic minority background.
29
Additionally, Table 3 sets out the congruence scores of the eight different voters’ categories and their policy preferences, based on the policy preferences of the representatives. The red color resembles the lowest congruence scores between voters and representatives. High scores of congruence are reflected by the green colors. The color-coding immediately makes it clear that when it comes to the issue of the influence of immigrants on the economy, the representatives do not share their preferences with the electorate. Conversely, the policy issue of ‘integration’ shows a high level of congruence, which means that in contrast to the aforementioned issue, the preferences of the representatives do correspond with the preferences of the voter.
TABLE 4. Congruence scores
Immigrants and economy Tax Expenditure Integration Military intervention Total Male White 67,8% 84,6% 75,5% 78,3% 306,2 Male Asian 56,5% 73% 75,8% 75,3% 280,6 Male Black 45% 74,8% 79,9% 75,3% 275 Male Mixed 57% 77,7% 81,9% 78,2% 294,8 Female White 74,3% 81,1% 75,1% 68,6% 299,1 Female Asian 58,4% 72,4% 80% 72,1% 282,9 Female Black 49,5% 72,6% 79,5% 73,8% 275,4 Female Mixed 67,3% 62,1% 84,3% 73,9% 287,6 Total 475,8 598,3 632 595,5
Note: The total values are color-coded ranging from red (low congruence) to green (high congruence).
Lastly, when Table 3 is being read horizontally, it reveals that both male and female respondents with a white British background are, as expected, the most congruent with their political representatives. The ‘second place’ is taken by the male and female respondents with a mixed background. The fact that those (men and women) with a black or Asian ethnic background are the most underrepresented is even more troublesome knowing that these form the two most dominant ethnic minority groups within the UK.
30
Overall, these results indicate that the representation of those with a background from an ethnic minority is significantly lower than the representation of someone with a ‘white British’ background. However, contrary to the expectations, this research did not find a significant difference between BAME women and BAME men. Therefore, the question whether or not political representation of policy preferences in the United Kingdom is biased against BAME women cannot be with a single yes or no. This study does confirm differences in how well the citizens are being represented; but the negative bias is aimed at both men and women with a BAME background, instead of only against women.
5. Conclusion
This research was undertaken to stress the need for a new approach to political representation, hence the introduction of ‘intersectional representation’; the idea that elected and non-elected representatives in democracies should represent the preferences, to the highest possible level of congruence, of all its citizens, emphasizing the inclusion of those with multiple dimensions of disempowerment. Put differently, it stresses the importance of having a collective body of representatives whose policy preferences accurately correspond to the preferences of all citizens, with the acknowledgment of individual intersections of race and gender. For the analysis of the combined effects of race and gender on the political representation of those who face multiple kinds of marginalization (women from an ethnic minority background) the United Kingdom has served as a case study. The key question addressed in this research was to what extent the policy preferences of ethnic minority women in the United Kingdom are congruent to the policy preferences of their representatives when compared to those with different intersections of race and gender. The findings of the data analysis bring us to the following conclusions in support of ‘intersectional representation’.
31
The first conclusion that can be drawn from the analysis, is that when groupings of those who have one or multiple dimensions of disempowerment based on gender or race are compared to the group that has none (‘white males’), one can see that almost all of these groups have significantly different policy preferences. As one might expect, there were significant differences found between ‘white males’ and all the other combinations of race and gender on the two topics of ‘integration’ and ‘the influence of immigrants on the economy’, meaning that these different intersectional groups prefer different takes on these two policy issues. However, no significant differences were found between the policy preferences of ‘white males’ when compared to men and women from a mixed ethno-racial background on the topics of ‘Tax expenditure’ and ‘Military intervention’. A possible explanation for this could be that those that considered themselves to belong to a ‘mixed’ background, have one parent who considers herself/himself to belong to the ‘white’ group.
Secondly, the conclusion drawn from the analysis of the policy preferences of politicians combined with those of the voters, categorized by their ethno-racial background, indicates that the representation of those with an ethno-racial background is significantly lower on all four policy issues. This conclusion can be drawn from the results of this analysis because there are significant differences between the level of congruence of respondents with a white background and respondents from an ethnic minority background. However, there were no highly significant differences found when the data were analyzed based on gender. This means that the differences in the level of representation between men and women, when compared to the preferences of the politicians on these four policy issues, are small.
The final analysis, which used the ‘intersectional’ many-to-many approach, leads us to the confirmation of the previous conclusion. When the levels of congruency are compared on the four policy issues, it provides us with additional support that white and mixed white citizens are better represented than those with an Asian or Black ethno-racial background, while at the
32
same time not enough support was found for the claim that women are less well-represented than other groups. However, the fact that no negative bias was found particularly against women in the United Kingdom, does not mean we should therefore immediately abandon intersectional representation and return to turning a blind eye to the people that face multiple dimensions of disempowerment. On the contrary, the absence of difference between men and women yet again emphasizes the need of ‘intersectional representation’, because if you want to evaluate the representation of all citizens, you will need to check for these differences for all possible combinations of personal dimensions.
The third conclusion that can be drawn from this research is that when it comes to the issue of the influence of immigrants on the economy, the representatives do not share their preferences with the electorate. This means that representatives are more inclined to think that immigrants have a good influence on the economy, while the voters tend to believe that immigrants do not have a good influence on the economy. Conversely, the policy issue of ‘integration’ shows a high level of congruence, which means that in contrast to the aforementioned issue, both the representatives and the voters take a similar position.
Lastly, it can be concluded from this analysis that both male and female respondents with a white British background are, as expected, the most congruent with their political representatives, while men and women from a black ethno-racial background show a lower level of congruence. Thus, these results indicate that the representation of those with a background from an ethnic minority is significantly lower than the representation of someone with a ‘white British’ background. However, contrary to the expectations, this research did not find a significant difference in policy preferences between BAME women and BAME men. Therefore, the question of whether or not political representation of policy preferences in the United Kingdom is biased against BAME women cannot be with a single yes or no. This study does confirm differences in how well the citizens are being represented; but the negative bias
33
is aimed at both men and women with a BAME background, instead of only against women. Additionally, as a little side note, the fact that those (men and women) with a black or Asian ethnic background are the most underrepresented is especially troublesome knowing that these already two already form the biggest ethnic minority groups within the United Kingdom and will only become more dominant in the next generations (Office for National Statistics, n.d).
This project provided an important opportunity to advance the new concept of ‘intersectional representation’. It offers some meaningful insights into the level of representation of those who face one or multiple dimension(s) of marginalization in the United Kingdom. However, the reader should bear in mind that the UK served as a most-likely case, which means that the UK is almost certain to fit the theory if the theory is true for any case at all (Levy, 2008: 12). This means that the generalizability of these results is subject to certain limitations. This is caused by the fact that it does not logically follow from this research, that in a case where underrepresentation of women from an ethnic minority background is less likely, the results of the current study will repeat itself.
Future research should, therefore, concentrate on the investigation of intersectional representation in different countries. This should be preferably undertaken within a least-likely context, to see what the difference in the representation of those who face multiple dimensions of marginalization is in a setting where mis- or underrepresentation is less likely to happen. This was not feasible for this research since the United Kingdom is the only country at present that specifically asks the respondents of the national surveys the question to which ethnic group they feel like they belong, which is needed for the empirical part of intersectional representation.
Furthermore, it would be interesting for future research on intersectional representation to assess the impact of having one parent belonging to a ‘white background’ in combination with having one parent from an ethnic minority background. The current analyses showed that
34
their policy preferences are more similar to those who felt like they belonged to a complete white background, compared to those from an ethnic minority background. Perhaps future research could analyze whether or not there are clear distinctions between the different combinations of ‘white background’ plus any other ethnic minority background.
In conclusion, whilst, unfortunately, this study was limited to four policy issues due to data constraints, it did give us valuable insight into the importance of the approach to representation as being ‘intersectional’. Without specifically looking at the representation of those who face multiple dimensions of disempowerment, you would risk ignoring the existing differences in the representation of these groups. It is not enough to look at gender and race as two parallel categories, because, even though this was not the case for the United Kingdom, their preferences could be distributed completely different. The ideal of having a legislature that accurately mirrors the policy preferences of its society is still at risk when governments are reduced to a mean and gender and race are treated as independent from each other. Looking at representation from an intersectional perspective urges us to see the differences between all the combinations of disempowerments which perhaps could finally lead to not only ‘record numbers’ of BAME MPs, but also a record number of intersectional congruence for all.
35
6. Bibliography
Alexander-Floyd, N. G. (2012). Disappearing Acts: Reclaiming Intersectionality in the Social Sciences in a Post-Black Feminist Era. Feminist Formations, 24(1), 1–25.
Andrews, K. (2019, 12 april). When it comes to Brexit, politicians are only respecting the ‘will’ of the white people. Geraadpleegd op 13 juni 2019, van
https://www.independent.co.uk/voices/brexit-extension-theresa-may-brussels-eu-race-bame-peoples-vote-a8865346.html
Ansolabehere, S., & Jones, P. E. (2011). Dyadic Representation. Oxford Handbooks Online, . https://doi.org/10.1093/oxfordhb/9780199559947.003.0013
Browning, S. (2019). Ethnic diversity in politics and public life (House of Commons Briefing Paper No. SN01156). Geraadpleegd van www.parliament.uk/
briefing-papers/sn01156.pdf
Bloemraad, I. (2013). Accessing the Corridors of Power: Puzzles and Pathways to
Understanding Minority Representation. West European Politics, 36(3), 652–670. https://doi.org/10.1080/01402382.2013.773733
Celis, K., Childs, S., & Curtin, J. (2016). Specialised Parliamentary Bodies and the Quality of Women's Substantive Representation: A Comparative Analysis of Belgium, United Kingdom and New Zealand. Parliamentary Affairs, 69(4), 812–829.
https://doi.org/10.1093/pa/gsw007
Cho, S., Crenshaw, K. W., & McCall, L. (2013). Toward a Field of Intersectionality Studies: Theory, Applications, and Praxis. Signs: Journal of Women in Culture and Society, 38(4), 785–810. https://doi.org/10.1086/669608
36
Chun, J. J., Lipsitz, G., & Shin, Y. (2013). Intersectionality as a Social Movement Strategy: Asian Immigrant Women Advocates. Signs: Journal of Women in Culture and Society, 38(4), 917–940. https://doi.org/10.1086/669575
Cooper, B. (2015). Intersectionality. Oxford Handbooks Online, . https://doi.org/10.1093/oxfordhb/9780199328581.013.20
Crenshaw, K. (1991). Mapping the Margins: Intersectionality, Identity Politics, and Violence against Women of Color. Stanford Law Review, 43(6), 1241.
https://doi.org/10.2307/1229039
Duncan, P., & Holder, J. (2017, 24 september). Revealed: Britain's most powerful elite is 97% white. Geraadpleegd op 13 juni 2019, van
https://www.theguardian.com/inequality/2017/sep/24/revealed-britains-most-powerful-elite-is-97-white
Fisher, S., Heath, A., Sanders, D., & Sobolewska, M. (2010). British Election Study Ethnic Minority Survey 2010 [Dataset]. Retrieved from
https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=6970#!/details FORS. (2016, 15 december). The Comparative Candidates Survey Module I - 2005-2013
[Dataset]. Retrieved on 13 mei 2019, van https://forsbase.unil.ch/project/study-public-overview/15760/0/
Golder, M., & Stramski, J. (2010). Ideological Congruence and Electoral Institutions. American Journal of Political Science, 54(1), 90–106.
Hague, R., & Harrop, M. (2013). Comparative Politics: Theory and Methods (9e ed.). Hampshire, United Kingdom: Macmillan.
37
Hancock, A. (2007a). Intersectionality as a Normative and Empirical Paradigm. Politics & Gender, 3(02). https://doi.org/10.1017/s1743923x07000062
Hancock, A. (2007b). When Multiplication Doesn't Equal Quick Addition: Examining Intersectionality as a Research Paradigm. Perspectives on Politics, 5(01). https://doi.org/10.1017/s1537592707070065
Hancock, A. (2013). Empirical Intersectionality: A Tale of Two Approaches. In O. Hankivsky, & J. S. Jordan-Zachery (Eds.), The Palgrave Handbook of Intersectionality in Public Policy (pp. 259–296). Retrieved from https://www.law.uci.edu/lawreview/vol3/no2/hancock.pdf
Hanretty, C., Lauderdale, B. E., & Vivyan, N. (2016). Dyadic Representation in a Westminster System. Legislative Studies Quarterly, 42(2), 235–267. https://doi.org/10.1111/lsq.12148
Hill Collins, P. (1990). Distinguishing features of black feminist thought. In P. Hill Collins (Red.), Black Feminist Thought: Knowledge, Consciousness, and the Politics of Empowerment (pp. 21–43). Boston, United States of America: Unwin Hyman. Kantola, J., & Nousiainen, K. (2009). Institutionalizing Intersectionality in Europe.
International Feminist Journal of Politics, 11(4), 459–477. https://doi.org/10.1080/14616740903237426
Kentish, B. (2017, 10 juni). Election results: Record number of black, Asian and ethnic minority MPs elected to parliament. Geraadpleegd op 12 juni 2019, van
https://www.independent.co.uk/news/uk/politics/election-results-bame-mps-elected-parliament-black-asian-ethnic-minority-commons-first-sikh-woman-a7783111.html
38
Levy, J. S. (2008). Case Studies: Types, Designs, and Logics of Inference. Conflict Management and Peace Science, 25(1), 1–18.
https://doi.org/10.1080/07388940701860318
Lombardo, E., & Verloo, M. (2009). Institutionalizing Intersectionality in the European Union? International Feminist Journal of Politics, 11(4), 478–495.
https://doi.org/10.1080/14616740903237442
Manin, B. (1997). The principles of representative government. Cambridge, United Kingdom: Cambridge University Press.
Mansbridge, J. (1999). Should Blacks Represent Blacks and Women Represent Women? A Contingent "Yes". The Journal of Politics, 61(3), 628–657.
https://doi.org/10.2307/2647821
Mansbridge, J. (2003). Rethinking Representation. American Political Science Review, 97(04), 515–528. https://doi.org/10.1017/s0003055403000856
Mansbridge, J. (2011). Clarifying the Concept of Representation. American Political Science Review, 105(03), 621–630. https://doi.org/10.1017/s0003055411000189
McCall, L. (2005). The Complexity of Intersectionality. Signs: Journal of Women in Culture and Society, 30(3), 1771–1800. https://doi.org/10.1086/426800
Mügge, L. M., & Erzeel, S. (2016). Double Jeopardy or Multiple Advantage? Intersectionality and Political Representation. Parliamentary Affairs, 69(3), 499–511.
https://doi.org/10.1093/pa/gsv059
Office for National Statistics. (n.d.). 2001 Census Statistics. Retrieved June 20, 2019, from https://www.nomisweb.co.uk/home/census2001.asp
