In this article, we elucidate crucial dimensions of feminist theories of representation through an empirical test of whether there is congruence between the political attitudes of women politicians and women citizens. A growing literature on political representation is driven by feminist concerns to assess the representation of women in different political settings. Seeking to understand contemporary political practice, feminist scholars have revisited representation theory to develop an extensive literature which addresses concerns that are not systematically treated in the mainstream political science literature on representation.Footnote 1 This feminist scholarship has enhanced our understanding of the nature of representation.Footnote 2 They find weaknesses and lacunae in established conceptions of representation that, inter alia, mask the exclusion of women and women’s interests from authoritative political deliberation and decision making. Much of this literature takes as its point of departure Hanna Pitkin’s well-established four-part typology of representation, which makes distinctions between formal, descriptive, substantive and symbolic representation.Footnote 3
Expressed in terms of Pitkin’s typology, feminists seek to explain the connections between descriptive (presence) and substantive representation. Pitkin described substantive representation as ‘acting in the interests of the represented in a manner responsive to them’.Footnote 4 Substantive representation is most commonly operationalized as ‘policy responsiveness’.Footnote 5 Women are thought to be represented when deliberations about public policy consider the potential impact on different groups of women. Descriptive representation occurs when representatives mirror the backgrounds of the represented.Footnote 6 There is not necessarily a relationship between descriptive and substantive representation, and some theorists reject the idea that there is any necessary connection between the two.Footnote 7 Recently, Jane Mansbridge has argued that the sex of a candidate does not matter so long as the interests of both men and women are represented in national parliaments;Footnote 8 however, the extensive practical requirements that she acknowledges are necessary for such representation are probably prohibitive in any complex society.
A common concern of feminist scholarship is the extent to which the interests of particular groups of voters (normally women) are represented in legislative decisions. Since Phillips argued so effectively for a ‘politics of presence’, feminists have contended that a necessary condition for the representation of women’s interests is the presence of women in decision-making bodies.Footnote 9 Phillips reasons that interests are realized in the course of deliberation and decision making, as various options, implementation strategies and competing concerns are discussed. Only when present may women benefit from such realization and insert their interests. While the logic of Phillip’s claim is inescapable, it has proved difficult to demonstrate that the representation of women’s interests necessarily follows from the presence of women representatives, although a great deal of circumstantial evidence that this is the case has been assembled and presented.Footnote 10 An important qualification to ‘presence effects’ is found in the interplay between party and sex. Many studies have found that party affiliation explains more than sex,Footnote 11 particularly in highly visible and confrontational settings. Moreover, it is very often women members of left-leaning parties rather than all women representatives who make the most effort to raise women’s issues and concerns. But such findings indicate that research on representation should take account of party differences, not that it should ignore differences between women and men. There remain significant differences between the sexes within parties.
Despite some reservations, most feminist political scientists are attracted to the potential of descriptive representation to deliver at least a measure of substantive representation. For example, in her account of six arguments that are commonly used to support the descriptive representation of women, Susan Dovi offers two arguments that support the case for women’s descriptive representation.Footnote 12 The transformative argument predicts that increasing the presence of women will transform politics by improving the democratic functioning of legislatures.Footnote 13 It assumes that women representatives will behave in a more democratic fashion and will pay more attention to political inequalities than men.Footnote 14 The ‘overlooked interests’ argument is that ‘male representatives are not always aware of how public policies affect female citizens’.Footnote 15 Descriptive representation is not achieved by the admission of a token woman. Thus, Laurel Weldon argues that the presence of individual or lone women is insufficient to guarantee the substantive representation of women because ‘individuals can rarely provide a complete account or analysis of the obstacles confronting the group without interacting with others from the group’.Footnote 16 Moreover, some circumstances particularly merit descriptive representation, which Jane Mansbridge argues can be justified in four contexts: contexts of mistrust; uncrystallized or not fully articulated interests; historical political subordination; and low de facto legitimacy.Footnote 17
In short, albeit with many caveats, the nature of the relationship between descriptive and substantive representation, as well as between women voters and women representatives, continues to preoccupy feminist scholars. This article addresses a central component of these feminist debates regarding the concept of representation: Phillips’s notion of uncrystallized interests. The argument is straightforward: if women at both the mass and elite level have a shared political interest that is not fully ‘crystallized’ into a coherent set of policy demands through public deliberation, then male representatives cannot unproblematically undertake the substantive representation of women.Footnote 18
We deduce from this literature that empirical research on the relationship between the attitudes of women voters and representatives must take account of four theoretical problems: (1) the nature of women’s interests, (2) the attendant danger of essentialism, (3) the mechanisms that connect presence to advocacy, and (4) the related difficulties of describing elite behaviour. We fulfil these four requirements by constructing an elite/mass comparison of women’s and men’s attitudes in Britain. We hypothesize that evidence of a commonality between the political attitudes of women at the mass and elite level is suggestive of uncrystallized interests; although women may not overtly express a shared sense of feminism or gender identity, they may, in fact, tacitly hold similar views or have an implicit feminist orientation. We define feminist orientation as a belief in equality between the sexes – combined with support for measures to improve gender relations – that may be found among both men and women; the concept is distinct from feminist consciousness.Footnote 19
Any account of the relationship between the descriptive and substantive representation of women requires the operationalization of a concept of ‘women’s’ interests, a longstanding problem for scholars of women and politics.Footnote 20 Women are a diverse group who may have less in common with each other than with similar groups of men. One way to circumvent the vexed problem of defining ‘objective’ interests is to focus instead on preferences of constituents as expressed in survey data. The reasoning here is that an alignment of political beliefs and values between women politicians and women in the electorate is a minimum requirement for ‘women to act for women’. So, while we cannot identify an objective women’s interest that differs from men’s, we can, and do, show that women’s self-stated preferences and attitudes differ from those of men.
Because women are not a homogeneous group, attempts to increase the descriptive representation of women must take account of variations among women. Failure to do so causes what Dovi calls an inclusion problem and amounts to an essentialist account of what it is to be a woman suggesting a binary and ‘natural’ division between the sexes.Footnote 21 Feminists are well aware that ‘[w]omen differ when they have children or do not, are divorced or not, have been raped or not, are straight or gay, obese or thin, Muslim or Christian, menopausal or prepubescent’.Footnote 22 Ignoring such differences may lead to misunderstandings of political attitudes and behaviour. Dovi argues that ‘feminists should not presume that privileged representatives – whether they are white males or white females – promote hierarchies, or that all representatives from disadvantaged groups seek to undermine hierarchies’.Footnote 23 The misunderstandings that arise from such assumptions are most likely to occur when dichotomous sex differences between women and men are used unaccompanied by any concept of gender.Footnote 24
In this research, we try to overcome some of the inclusion problems by taking into account variations between women and men, and testing whether the differences we find between them are consistent across different social categories. By examining those differences between women and men that our data permit, we attend to some of the group identities that are implicated in gender differences. Retaining the dichotomous concept of sex, we also keep in play the differences between different groups of women and men. The survey data that best suit our purposes do not include information about victims of crime and respondent’s sexuality or weight, but we can consider some important differences between women and men, such as occupational class, marital status, religiosity, place of residence and age. It is unlikely that we would find that all men support the status quo whilst all women wish to challenge it. Instead, we hypothesize that more women than men (across a variety of social groups) favour greater equality for women, but we expect to find a number of feminist men and anti-feminist women. Our comparison of elite/mass attitudes addresses this issue by examining average responses. In so far as women and men have different preferences, increasing the number of women in parliament should increase the average presence of attitudes favourable to women’s preferences. In short, we avoid making essentialist claims about all men and all women by considering differences within each group and reflecting these both by calculating average responses and by using controls for group differences.
An important argument for women’s presence in legislatures is that details matter and that processes of deliberation lead to the articulation of previously uncrystallized interests.Footnote 25 Our research design enables us to test for evidence of uncrystallized interests through comparison of attitudes of various groups of voters and representatives. Finally, we overcome the problems of data collection associated with investigating elite behaviour, which are often hidden behind the scenes, by examining the elite attitudes that are a prerequisite for behaviour.Footnote 26 Therefore, this article meets all of the four requirements we have identified as necessary for empirical research on women’s representation.
Elite/Mass Comparison
Our research contributes to the elite/mass debate by considering whether there is a congruence in the political attitudes of women politicians and women citizens. We draw on and contribute to the political science of elite/mass attitude comparison.Footnote 27 Phillip Converse tested what he described as ‘issue constraints’;Footnote 28 he compared inter-correlations on issues assuming that high correlation coefficients supported the notion that a single liberal–conservative scale underlay attitude structures. He found a close relationship between different issue preferences in political elites, but no such strong correlation was evident in the mass public. His research provoked an ongoing debate about both the existence of ideological consistency in attitudes and beliefs at the mass level and also the fit of mass attitudes with the attitudes of political elites.Footnote 29 Research has established at least two dimensions to mass political attitudes and considerable evidence of internal consistency within the left–right and liberal–authoritarian scales.Footnote 30 The debate about the structure and consistency of political attitudes has responded to claims that post-materialist values have gained ground among the mass public.Footnote 31 Inglehart’s post-materialist theory claims that the generations born since the Second World War have enjoyed more financial security than previous generations and have therefore begun to prioritize non-material political issues, such as the environment, civil liberties and individual improvement.Footnote 32 Post-materialist values have been associated with feminist orientations.Footnote 33 Rather less research has been undertaken into the attitudes of the British public to equality between women and men. Moreover, there are few studies that either focus on Britain or include it in comparative analyses.Footnote 34 Britain is not among the main cases that Norris and Inglehart analyse in their study of sex equality and cultural change around the world.Footnote 35
A few authors have examined aspects of elite or mass attitudes to women’s equality in Britain. Bernadette Hayes tested whether feminist orientations amongst the electorate predicted Labour party votes at the 1992 general election.Footnote 36 Hayes operationalizes feminist orientations as support for women’s rights or equal opportunities between the sexes. Hayes found that ‘just over half of the male respondents ‘claimed that equal opportunities for women in Britain had not gone far enough [compared with] 60 per cent of women [who] endorsed this position’.Footnote 37 Using data from the 1997 general election, Rosie Campbell established that women born after 1944 are more feminist oriented than either men or older women.Footnote 38 But we lack research that looks at the change in attitudes over time. At the elite level, research by Joni Lovenduski and Pippa Norris demonstrated that women candidates and Members of Parliament (MPs) are more likely than men candidates and MPs to support a range of equality measures designed to increase the number of women representatives.Footnote 39 However, elite and mass attitudes on these issues have not previously been compared in a single British study.Footnote 40
The mainstream empirical representation literature provides a variety of measures of the ‘representativeness’ of legislatures. In studies of the United States, research correlates the behaviour of individual legislators with those of their constituents; for example, exploring whether legislators tend to behave as trustees or delegates.Footnote 41 This approach is not viable in the British case for two reasons. First, large-scale studies contain too little data at the constituency level for reliable comparison of the average constituent’s opinions to those of their representatives. Secondly, it is inappropriate in contexts of strong party discipline.Footnote 42 Therefore, we investigate whether there is any congruence between the average attitudes of groups of citizens and members of political elites. Such congruence is a prerequisite for representation but not a guarantor of it. An alternative mode of analysis is to compare mass opinion to changes in government policy.Footnote 43 While such comparison might provide some interesting insights, it would not help us address the relationship between the descriptive and substantive representation of women, simply because too few women have held decision-making positions in government to assess the relationship between their presence in office and variations in gender-equality outcomes.
Accordingly, here we focus specifically on the representativeness of the British political elite on gender-equality issues. We argue that gender equality is central to any reasonable construction of democracy and is very often left out of the range of attitudes used in mainstream studies of representation, which typically focus on comparison of elite and mass left–right positions,Footnote 44 or on attitudes to social welfare, foreign policy and civil rights.Footnote 45 We do not, however, make any claim of being able to assess the overall or general representativeness of the British political elite.
Data And Methods
The guiding research question of this study is whether women in the political elite share political attitudes with women in the mass public. We use the British Election Study (BES)Footnote 46 series to measure mass attitudes and the British Representation Study (BRS) series for elite attitudes.Footnote 47 The BRS is a series that surveys all candidates and sitting MPs at each general election. The series does not include exactly the same questions or wording for all of the items used in the BES analysis, but many similar measures are included: hence, comparison is appropriate. In order to operationalize our research question, we first identify an issue area for comparison. We conduct our analysis on attitudes to equality between the sexes, because it is clearly an issue that affects men and women differently and, as such, is an area where the descriptive representation of women may impact directly on their substantive representation. Although the BRS and BES surveys do not carry the same gender equality questions and the BES does not include the same questions in each survey, they each include a range of items that allow us to attempt to answer our research question.
Over the years, the BES has asked different questions about the equality of women and men. Hence, it is not possible to trace attitude change through responses to the same question election after election. The question most frequently used throughout the series asks how respondents feel about ‘attempts to give equality opportunities for women in Britain’, and was included in the 1974 (October), 1979, 1987, 1992 and 1997 BESs.Footnote 48 We hypothesize that women will be more likely than men to think that such attempts to achieve equality have not gone far enough and more likely to support the strongest measures to achieve equality. We also assess differences between age cohorts because research has shown that younger and older women hold different political opinions.Footnote 49 As well as including the usual range of background characteristics (class, marital status and education) as controls, party of vote and urban/rural location were added. Party of vote is a crucial element of the elite/mass comparison as party loyalty is likely to be a strong predictor of attitudes at the elite level. Urban/rural location is also an important factor because it has been shown elsewhere to have a significant impact upon feminist attitudes.Footnote 50 In addition, we consider the small number of other gender-related questions that are asked twice in the BES survey series, in the 2001 and 2005 BESs. This repeated battery is useful because, using factor analysis, we can explore whether all of the items are best explained by the same underlying phenomena. After the factor analysis, the scores for any important underlying factors are saved and used as dependent variables in subsequent analysis.
Using a combination of ordinary least squares (OLS) and ordinal regression analysis,Footnote 51 we test whether men and women do have different attitudes to their equality with each other, adding controls for other background characteristics including age, class, education, religiosity, marital status and party preference on attitudes to traditional gender roles and to the descriptive representation of women. We then repeat the analysis for the elite attitudes using the British Representation Studies 2001 and 2005, allowing us to compare the two sets of results to establish whether the attitudes of women in both mass and elite publics are congruent.Footnote 52
Mass Attitudes To Equality Between Women And Men
Accordingly, we first assess changes in mass attitudes to gender equality over time. As explained above, the most frequently available suitable question in the BES series was asked in the 1974 (October), 1979, 1987, 1992 and 1997 studies and asks how respondents feel about ‘attempts to give equal opportunities for women in Britain’ (responses range from ‘gone much too far’ to ‘gone not nearly far enough’).Footnote 53 The datasets for these years were combined to create a new dataset with 14,745 valid responses to the ‘gone too far’ question.Footnote 54 The amalgamated dataset allows us to look at change across time and birth cohorts.
Figure 1 describes a small emerging sex difference. Men and women respondents born before 1945 had similar attitudes to gender equality, except in the oldest cohort where women were slightly more likely than men to feel that equal opportunities for women had gone too far. In contrast, in the younger cohorts, women were slightly more likely than men to say that it had not gone far enough.Footnote 55 Respondents’ attitudes show no perfect linear trend, but overall the number of respondents reporting that equality between the sexes had gone too far reduced from 19.2 per cent in 1974 to 9.2 per cent in 1997. The number of respondents who felt it had not gone far enough grew from 35 per cent in 1974 to 41 per cent in 1997. Further analysis shows that the members of the older cohorts remained more conservative in their views about equality for women even in the later election years. Thus, we see three patterns: (1) younger generations are more likely than older generations to feel that equal opportunities for women have not gone far enough; (2) the electorate has become slightly more likely to think that equal opportunities have not gone far enough over time; and (3) there is a divergence between men and women in younger cohorts, where women are now more likely than men to say that equal opportunities have not gone far enough. Thus, our hypothesis that women are more in favour of gender equality than men is confirmed, with differences mainly reflecting the greater feminist orientation of younger women.
Fig. 1 Average responses to the question on how respondents feel about ‘attempts to give equality opportunities for women in Britain’, from the 1974 (October), 1979, 1992 and 1997 BESs Note: The difference between the sexes is statistically significant at the 0.001 level in the 1905–14, 1945–54, 1955–64, 1965–74 and 1975–84 birth cohorts.
In order to test the robustness of these three patterns, we undertake simple ordinal regression analysis, presented in Table 1.Footnote 56 It shows that, even after controlling for a number of other background factors, women are more likely than men to think that equality between the sexes has not gone far enough. Birth cohort and election year both had a significant independent effect, with younger respondents and respondents in later elections more likely to think that equality for women has not gone far enough. As expected, in terms of education, throughout the surveys, respondents with degrees are more likely to think that equality for women has not gone far enough. People looking after the home (in earlier surveys, these people were described as housewives and were exclusively women), people who regularly attend religious services and Conservative party voters are all more conservative about equal opportunities than other respondents.Footnote 57 This analysis indicates that, overall, women are more likely than men to think that equal opportunities for women have not gone far enough, a difference that is greatest amongst younger generations.
Table 1 Ordinal Regression on Attitudes to Equal Opportunities for Women, British Election Studies, 1974 (October), 1979, 1987, 1992 and 1997
Notes: N = 10,831. Equation chi square 512 (df 8). Equation −2 log likelihood 11,587.
There was no evidence of multicollinearity between the independent variables.
***Significant at the 0.001 level.
Having used the ‘not gone far enough’ question to assess change over time and across groups, we can use the battery of items that are available in the 2001 and 2005 BESs to examine attitudes to gender equality and to women’s representation in more detail. Simple descriptive analysis is undertaken for each of the items before conducting a factor analysis to establish whether responses are best explained by one underlying phenomenon.Footnote 58
In both the 2001 and 2005 British Election Studies, respondents were asked to agree or disagree with the following statements:
— Women are better representatives of women’s interests than men
— Most men are better suited to politics than are most women
— Women need to get more involved in politics to solve problems that concern them
— A husband’s job is to earn the money; a wife’s job is to look after the home and family.
In order to proceed with the analysis, we recoded responses to the four statements so that all of the items ran in the same direction, giving those we conceived as a more feminist response the higher value.Footnote 59 An exploratory factor analysis was then undertaken to examine the relationships between the variables.Footnote 60 The analysis produced two factors with eigenvalues over 1.Footnote 61 The rotated solution suggested that the first factor was best explained by the two items that measured attitudes towards traditional (unequal) gender roles: ‘husbands should earn the money while the wife looks after the home’ and ‘men are better suited to politics’. The second factor was better explained by the two items that measure attitudes to the political representation of women: ‘women should be more involved in politics’ and ‘women MPs better represent women’. We use this factor to describe attitudes towards the descriptive representation of women because the two measures refer directly to the presence of women in politics.Footnote 62 The factor scores were saved and used as dependent variables in later analysis.
The factor scores presented in Figure 2 move from support for traditional gender roles to hostility to them. We see a sharp increase in hostility to traditional gender roles across the older generations that levels off at a relatively high level of support for equal roles among the younger generations. Throughout the different birth cohorts, women are slightly more hostile to traditional roles than men, a difference that is most pronounced in the younger generations. Here is further evidence of a divergence between younger men and women, with younger women becoming more supportive of equal gender roles than younger men.
Fig. 2 Factor scores for hostility to traditional gender roles by sex and birth cohort, British Election Studies, 2001 and 2005
Mass Attitudes To Women’s Representation
Figure 3 shows the relationship between attitudes to the descriptive representation of women by sex and birth cohort. Overall, members of older generations of both sexes are more likely to think that women MPs represent women and that women should be more involved in politics, with older women more supportive than older men. Younger generations are less likely to hold these views. However, in the younger generations, differences between the sexes are larger than in the older cohorts. The support of men for the descriptive representation of women is considerably lower than that of women. Hence, the pattern is of an emerging and growing sex difference, with younger women seeing more benefit in the descriptive representation of women than younger men.
Fig. 3 Factor scores for attitudes to the descriptive representation of women by sex and birth cohort, British Election Studies, 2001 and 2005
Paradoxically, we find a negative correlation between liberal attitudes to equal gender roles and support for the descriptive representation of women. As generations have come to be more hostile to traditional gender roles, their support for the descriptive representation of women has declined. Although they are more hostile to traditional gender roles than younger men, younger women are less likely than older women to think that women need women MPs. We did not expect to find this pattern at the outset, but it makes intuitive sense in that individuals who favour equality may disagree with policies to increase women’s representation; they are likely to believe that women can represent men and vice versa.
To explore further the relationships between sex and attitudes to equality between women and men, we conducted a regression analysis using the factor scores as dependent variables. The results of the analysis are presented in Table 2. The regressions were conducted using the 2005 BES only because it includes a number of measures of local difference generated using census and other data.Footnote 63
Table 2 OLS regression of hostility to traditional gender roles, British Election Study, 2005
***Statistically significant at the 0.001 level. †An interaction term.
The significant coefficient for sex in Model 1 of Table 2 shows that sex has an impact on attitudes to traditional gender roles. However, when a dummy variable indicating women born since 1945 is included in Models 2 and 3, the overall sex difference disappears. Thus, the sex difference in Model 1 is explained by the relatively greater hostility to traditional gender roles of younger women. The regression analysis was repeated and the independent variables were included one at a time in an attempt to isolate which particular factors reduced the sex/generation difference. This is not a precise procedure but helps to indicate which variables can begin to explain the sex difference. The inclusion of respondent’s Goldthorpe/Heath class reduces the coefficient from 0.199 to 0.148. We see, therefore, that the younger women across the board and professional women, more than similar men, are hostile to traditional gender roles.
Table 3 confirms that the relationship between sex and attitudes to the descriptive representation of women differs from that between sex and hostility to traditional gender roles. The sex difference in hostility to traditional gender roles was explained by the attitudes of younger women, particularly professional younger women. But when we consider attitudes to the descriptive representation of women, we find that the inclusion of a range of control variables does not eliminate the overall effect of sex identified in Model 1, and still evident in Models 2 and 3. Overall, older people and women are more in favour of the descriptive representation of women than are younger people or men. There is a weaker but statistically significant relationship between attitudes to the descriptive representation of women and urban/rural location, with respondents from more densely populated areas most in favour.Footnote 64 The combination of these two trends means that there is a further divergence between the opinions of younger women and men, with support for the descriptive representation of women falling less sharply across generations of women than men.
Table 3 OLS regression of attitudes towards the descriptive representation of women, British Election Study, 2005
***Statistically significant at the 0.001 level. **Statistically significant at the 0.01 level. *Statistically significant at the 0.05 level. †An interaction term.
Party of vote has a significant effect on attitudes to gender equality, to traditional gender roles and to the descriptive representation of women. Overall, Conservative voters were less likely than other voters to think that equality between the sexes has not gone far enough; they are also less likely to be hostile to traditional gender roles and less likely to support the descriptive representation of women.
The analysis presented so far shows some emerging divergence between men’s and women’s attitudes to gender equality in the mass public. Younger women are more likely to be hostile to traditional gender roles and to support the descriptive representation of women than younger men. There are also differences between women. Younger women are less likely than older women to be concerned about the descriptive representation of women, even though they remain more likely to support it than younger men. However, younger women in the electorate may not be aware that they are, on average, more concerned about equality between the sexes than men.
Elite Attitudes To Equality Between Women And Men
In the analysis of mass attitudes, we find potential for the misrepresentation of women’s views on gender equality. If political representatives are disproportionately men (and older men at that), and if their attitudes reflect those of men in the mass public, then the attitudes to gender equality voiced in our political institutions are likely to be biased in favour of traditional values that have less support among women. In that case, it is logical to argue that women might need women representatives to act for them in respect of gender equality, even if some women say they do not want them. To assess this proposition, we now consider political elite attitudes to gender equality. We use the British Representation Study (BRS) to establish whether the difference between the attitudes of women and men found in the British mass public is also evident in the political elite. Because the BRS series cannot be simply combined to create a larger dataset, we analyse each survey separately.Footnote 65
Using the BRS, it is possible to make a more fine-grained analysis of attitudes to gender equality than with the BES. The BRS includes separate measures of attitudes to the descriptive representation of women that capture how far respondents are prepared to go with strategies to increase the representation of women as well as measures of attitudes to traditional gender roles. Following Lovenduski, we measure support for three types of strategy to increase the presence of women in parliament: equality rhetoric, equality promotion and equality guarantees.Footnote 66 Equality rhetoric is the public acceptance of women’s claims and is measured by general attitudes to the descriptive representation of women. It is found in party campaign platforms and party political discourses and in the speeches and writings of political leaders. Equality rhetoric can affect the attitudes and beliefs of both prospective candidates and party selectorates. Equality promotion attempts to bring women into political competition by offering special training, financial assistance, setting targets for women at various stages of recruitment and adopting other measures that enable women to come forward. Equality guarantees are the requirements made by parties and governments to secure places for women representatives by making their sex a necessary qualification for office. Quotas, whereby places on electoral lists or particular constituencies or representative bodies are reserved for women, are examples of equality guarantees.Footnote 67
Preferences for more women representatives are an indicator of general support for descriptive representation. However, support for the practical measures to bring such increases, such as quotas or all-women shortlists, provides an indicator of the intensity of the preference measured by the strength of the mechanism that is supported. Equality rhetoric is the weakest, equality guarantees or positive discrimination the strongest, with equality promotion or positive action in between.Footnote 68 The analysis proceeds by looking at each component of attitudes to gender equality in turn.
Following this logic, we divide questions in the 2001 and 2005 BRSs into four components:
— Attitudes to traditional gender roles: measured by attitudes toward women’s role, attempts to give equal opportunities to women in Britain, responses to the statements: ‘government should make sure that women have an equal chance to succeed’; ‘most men are better suited emotionally to politics than most women’; ‘all in all, family life suffers when the woman has a full-time job’; ‘being a housewife is just as fulfilling as working for pay’ and ‘a husband’s job is to earn money; a wife’s job is to look after the home and family.’
— Attitudes to equality rhetoric: are measured by items on general attitudes to the descriptive representation of women. We therefore use the item ‘should Parliament have more women MPs’, which is also comparable to the measures of attitudes to descriptive representation identified in the 2001 and 2005 BESs.
— Attitudes to equality promotion: measured by attitudes toward party training programmes for women and financial incentives for women candidates.
— Attitudes to equality guarantees: measured by attitudes toward all-women shortlists, quotas or compulsory minimum numbers of women, reserved seats for women and responses to the statement ‘no more than 55 per cent of either sex should be elected’.
Elite Attitudes To Traditional Gender Roles
First, we consider the items that measure attitudes to traditional gender roles. Across all of the parties, women MPs and candidates are significantly more likely than men to disagree strongly with the statement, ‘most men are better suited emotionally to politics than most women’. Women across the parties are also more likely than men to disagree strongly with the statement, ‘all in all, family life suffers when the woman has a full-time job’. Women are less likely to think that, ‘being a housewife is just as fulfilling as working for pay’ in every group except for the Conservatives in 2001, where women were more likely to disagree with the statement, but the difference was not statistically significant. Finally, women disagree more strongly with the statement, ‘a husband’s job is to earn money; a wife’s job is to look after the home and family’ across all of the parties and the differences between the sexes are statistically significant. Thus, both surveys of political elite opinion in Britain indicate that women are more hostile to traditional gender roles than men.
The seven items relating to traditional gender roles were included in a factor analysis. In both the 2001 and 2005 BRSs there was one factor with an eigenvalue above 1; the factors explained 45 per cent and 40 per cent, respectively, of the variance. The resulting scales were sufficiently reliable with Cronbach’s alphas of 0.75 in 2001 and 0.69 in 2005. The factor scores are displayed by birth cohort and sex in Figure 4. The sex difference, with women more hostile to traditional gender roles than men, is statistically significant,Footnote 69 but the age differences that are suggested by the figure are not statistically significant. Thus, within the political elite, unlike the mass public, age or birth cohort does not seem to have an impact on attitudes to traditional gender roles.
Fig. 4 Hostility to traditional gender roles by sex and birth cohort, British Representation Study, 2001 and 2005
In order to test the descriptive analysis more rigorously, an OLS regression on the factor scores is conducted and presented in Models 1 and 5 of Table 4. Sex has a significant impact upon elites’ attitudes to traditional gender roles even after controlling for party, birth cohort, marital status, education and religiosity. Birth cohort had a weak but significant influence on attitudes to traditional gender roles in 2001 but not in 2005. Thus, it seems that age effects are more pronounced in the general population than in the political elite. As expected at the elite level, the standardized coefficients suggest that party is a more important predictor of attitudes than sex, but the sex coefficient is still significant after controlling for party and it is the second most important predictor. Both the Conservative and Liberal Democrat candidates and MPs were less hostile to traditional gender roles than Labour candidates and MPs, with the Conservative respondents least hostile of all. More religious members of the political elite were also less hostile to traditional gender roles.
Table 4 Regression Analysis of Attitudes towards Traditional Gender Roles and the Descriptive Representation of Women, British Representation Studies, 2001 and 2005
Notes: Models 2 and 6 are ordinal regressions; the reported R 2 for these models is the Cox and Snell pseudo; all other models are OLS regressions. Model 2: equation chi square 217 (df 7), equation −2 log likelihood 655. The threshold estimates with standard errors in parenthesis were 1.909*** (0.569), 3.877*** (0.584), 5.946*** (0.650). Model 6: equation chi square 78 (df 7), equation −2 log likelihood 612. The threshold estimates with standard errors in parentheses were 1.398* (0.678), 3.24*** (0.690), 5.528*** (0.788). ***Significant at the 0.001 level. **Significant at the 0.01 level. * Significant at the 0.05 level.
Elite Attitudes To Women’s Descriptive Representation
Equality Rhetoric
We take general attitudes to the descriptive representation of women as indicative of support for equality rhetoric. It is measured in the 2001 and 2005 BRSs by asking respondents whether ‘there should be more women MPs’. Women were significantly more likely to agree with the statement than men in each election year, with at least 20 per cent more women than men strongly agreeing. Birth cohort has no effect on general attitudes to the descriptive representation of women in 2001 but there is a significant relationship in 2005 with younger respondents less supportive, mirroring the mass level findings. The sex difference is also apparent within the responses from the Labour and Conservative parties,Footnote 70 while in the Liberal Democrats it is found only in the 2005 survey. In each party, women are more likely than men to feel strongly that there should be more women MPs, but these differences do not always reach statistical significance. In order to investigate these relationships further, an ordinal regression was conducted on the general descriptive representation/equality rhetoric measure and the results are presented in Models 2 and 6 of Table 4. Overall, sex and party are the most important factors with birth cohort significant in 2001 but not 2005. Again, we see evidence that the views of women in the political elite differ from those of men even after controlling for party.
Equality Promotion
The items relating to financial incentives and training programmes for women candidates measure attitudes to equality promotion.Footnote 71 Overall, women were significantly more in favour of equality promotion than men.Footnote 72 The only statistically significant difference between the sexes in their attitudes to party training for women within the parties occurs in the Conservative party – women are more strongly in favour than men. There are significant sex differences between men and women in their attitudes to financial incentives for women candidates, with more women in favour in each group in all of the parties. The descriptive statistics are supplemented with a regression analysis presented in Models 3 and 7 of Table 4. Sex has a significant effect after controlling for party, and is the second most important predictor of support for equality promotion. Party is the most important explanation but does not drown out the sex effect. Birth cohort had a weak impact on attitudes in 2001, and no impact in 2005. Thus, again we see that the sex effect evident in the mass public is also evident in the political elite, but the generation effect is less apparent.
Equality Guarantees
Descriptive analyses of the items relating to equality guarantees show that there are strong sex and party differences in attitudes to all women shortlists, quotas and reserved seats for women. The difference between the parties is stronger than that between the sexes, with members of the Labour party considerably more likely to be in favour of quotas than members of the other parties. Within the parties, there are clear sex differences. Women candidates and MPs from all of the parties are significantly more in favour of equality guarantees than men in every group except for the Conservatives in 2001, where women were marginally, rather than significantly, more in favour. It is worth noting that not a single male Conservative respondent strongly approved of quotas or compulsory minimum numbers of women in either 2001 or 2005.
Attitudes to Equality Strategies
The four items measuring equality guarantees, attitudes to quotas, all women shortlists, reserved seats for women and responses to the statement that ‘no more than 55 per cent of either sex should be elected’ were entered into a factor analysis, and the factor scores were saved as dependent variables. In both the 2001 and 2005 BRSs, one factor with an eigenvalue greater than 1 was identified. In both cases, more than 70 per cent of the variance was explained by the factor. The relationships between attitudes to equality guarantees and age and sex are presented in Figure 5. Overall, the sex difference is statistically significant, as is the birth cohort difference in 2005. This is the only elite level birth cohort effect that strongly mirrors the mass level attitudes to descriptive representation. At the mass level, younger generations are more hostile than older generations to equality rhetoric. At the elite level, we are able to investigate equality rhetoric, promotion and guarantees and find a pronounced difference between the age cohorts only in the 2005 BRS, where younger respondents are more hostile to equality guarantees than older respondents. These relationships are analysed further by using the factor scores as dependent variables in OLS regression; the results are presented in Models 4 and 8 of Table 4. The regression analysis confirms a significant generational effect in 2005, when younger members of the political elite were more hostile to equality guarantees than older members. The sex effect remains significant, but in this instance is the third most important explanation of attitudes with both Conservative and Liberal Democrat respondents significantly more hostile than Labour respondents. It would seem that Labour and Liberal Democrat members of the political elite share views on equality rhetoric and promotion but not on equality guarantees.
Fig. 5 Attitudes to equality guarantees by birth cohort and sex, British Representation Studies, 2001 and 2005
Discussion
Overall, our analysis suggests that sex differences in attitudes to gender equality at the elite level mirror the sex differences evident in the mass public. Women are on average more hostile to traditional gender roles and more supportive of measures to improve the descriptive representation of women. However, there are important distinctions between the two levels; generation has a more profound effect on the attitudes of the mass public, while party has a more profound effect on the attitudes of the political elite. This makes sense when one considers the elite/mass literature, which has focused upon the coherence of political ideology. Here, members of political elites are more likely to have consistent ideological positions. The relative consistency of ideology at the elite level is likely to produce a greater polarization by party and leave less room for generation effects.Footnote 73 In the mass public, there is a clear generation effect with members of older cohorts more supportive of equality rhetoric and less hostile to traditional gender roles. The generation effects in the political elite are not consistent. Party has a weak to moderate effect on political attitudes in the mass public. In the mass public, a Conservative vote is a better predictor of attitudes to traditional gender roles than sex, but age is the most important predictor, whereas mass attitudes to general descriptive representation or equality rhetoric are best explained by respondent’s age followed by sex, urban/rural location and then party. In the political elite, Conservative party membership is always the most powerful explanation of attitudes to gender equality, followed by sex, with the exception of attitudes to equality guarantees, where both Conservative and Liberal Democrat party membership trumps sex. These differences demonstrate that party is the best predictor of attitudes to gender equality, but within this the sex of the candidate matters. Like women in the electorate, women members of the political elite – in all of the parties – are more likely than men to be hostile to traditional gender roles and to be more supportive of measures that aim at increasing the descriptive representation of women than men.
Conclusions
The two trends we have identified in the mass public, of declining support for descriptive representation across generations accompanied by increased hostility to traditional sex roles in younger birth cohorts, are arguably moving in different directions.Footnote 74 The decline in men’s support for the descriptive representation of women has occurred more rapidly than women’s, while women’s hostility to traditional gender roles has risen more quickly than men’s. Thus, both trends point towards an increasing divergence between the attitudes of men and women in Britain. Furthermore, these differences between the sexes are evident in the political elite, both overall and within political parties.
The sex differences in attitudes that we report here take forward the argument about the relationship between the substantive and descriptive representation of women. In terms of attitudes to traditional gender roles, the trends in elite and mass attitudes reveal both similarities and differences between voters and representatives. The similarities and differences are both important. If men and women have different attitudes at the mass level that are reproduced among political elites, then the numerical under-representation of women might, all other things being equal, have negative implications for women’s substantive representation.Footnote 75 Put simply, in terms of attitudes to traditional gender roles, on average men and women differ, and women representatives are more like women voters and male representatives are more like male voters. Such differences diminish but continue when we control for age and party. On average, male representatives, and would-be representatives, do not report the same level of hostility to traditional gender roles as women voters; hence, they may, on average, be less likely than women representatives to act for, or otherwise represent, women voters on the many issues affected by such roles.
Attitudes to the descriptive representation of women present more of a puzzle. We may wonder why younger women are more hostile than older women to traditional gender roles but remain less in favour of the descriptive representation of women. There are three possible explanations. First, younger women who are hostile to traditional gender roles may not acknowledge the difficulties women face in political recruitment.Footnote 76 Secondly, younger women may (mistakenly) believe that their hostility to traditional gender roles is shared by their male peers. Thirdly, younger women may not appreciate that their hostility to traditional gender roles is not shared, to the same extent, by men at the elite level, the very men who are disproportionately present in British political institutions.
In showing that women and men have different attitudes, we contend that sex difference signals attitudinal differences, albeit mediated by other group characteristics. Accordingly, we challenge conceptions of representation that argue that the sex of the representative does not matter and that representation depends upon ideas but not the sex or gender of the representative. In the current state of gender relations in Britain, at least, women and men on average have different ideas about the roles of each sex and these might be described as uncrystallized or not fully conscious interests, with potentially considerable consequences for substantive political representation. In short, while women may not want more women representatives, they continue to need them.
