SOME GROUND CLEARING

The British Candidate Study was a joint project, and the research design in particular was the result of a very long and intensive process of discussion and development mostly between Pippa Norris and me, but also with some advice from academics and party officials. Probably the research would not have been done had it not been for that cooperation but, arguably, each of us would have done it differently on our own. I am sure that Pippa Norris would write this account differently.

This article has, of course, been written with the benefit of hindsight illuminated by the knowledge and understanding gained over the decades of research on gender and politics after the British Candidate Study. More important than hindsight to my current understanding of candidate selection politics are the subsequent theoretical developments in feminist political science which in the form of feminist institutionalism have become a highly articulated convincing approach to the study of political power in institutions.Footnote ³ Although not always acknowledged or well understood, our work informed these developments. One thing that always strikes me about the subsequent discussions of the version of the supply and demand model presented in Political Recruitment (Norris and Lovenduski Reference Norris and Lovenduski1995) is that they either tend to glide over the details of the research design and/or to ignore its flexibility. We were criticized by established political scientists for being too feminist, but that was only to be expected. Ironically, we are often criticized by feminists for being insufficiently feminist, for having only a biological concept of gender, for ignoring institutionalism, for oversimplifying our results and for being a-theoretical.

I dispute this. We have frequently been cited as stating that supply problems explained the under-representation of women when in fact we argued that both supply- and demand-side explanations needed to be considered. We were interested in providing a full account of legislative recruitment that spoke both to feminist scholarship and to political scientists. We aimed to write women into the script without trashing the valuable scholarship that came before. We responded to feminist concerns that there were so few women in elected office in Britain and also to long-standing preoccupations with class and race whereby working-class and minority ethnic people were also under-represented, a situation that affected women as well as men. Moreover, although we did not claim to be institutionalists, the British Candidate Study was very much a new institutionalist type of project. It was a carefully designed process-tracing study that aimed to answer the question of why white middle-class men were so much more likely than women and other men to be selected as candidates by political parties.

The development of the research design pre-dated the new institutionalist industry, but the elements of the approach were there. We used a multi-method design that combined observations, expert, official and candidate interviews, and survey work with careful explication of formal and informal rules, practices and processes (see Appendix). Far from treating gender as a simple matter of binary sex difference, we considered carefully its interaction with race and class. In fact the book is subtitled ‘Gender, Race and Class in the British Parliament’. In short, although Political Recruitment was a product of its time and undoubtedly left considerable room for development and refinement, it was an explicitly theorized, gendered, feminist, institutionalist multi-method examination of the practices that excluded women and most men from political representation roles in Britain. I am not arguing that there was nothing wrong with our work, rather complaining that so many of those who cite it so critically do not seem to have had the time to read it.

Such misrepresentation is more than an irritant. It is symptomatic of serious flaws in the way research is now packaged for publication, especially in scholarly journals where truncated length and citation requirements too often mean that important caveats, acknowledgements and explanations are cut from submissions by scholars working to unrealistic deadlines and ‘productivity’ requirements. Thus some of the misrepresentation of our scholarship is probably not deliberate but instead an inevitable product of a type of ‘fast science’ that is all too prevalent in contemporary academic writing. It arises from a combination of pressures to publish and scholarly conventions that favour claims to be ‘refuting’ or overturning previous scholarship over extensions, refinements and developments of previous work. A good example in feminist political science is found in the long argument about ‘critical mass’. In 1988 Drude Dahlerup published an article that applied Rosabeth Moss Kanter’s concept of critical mass (adapted from physics, possibly via the sociology of race relations) to an assessment of the impact of the growing presence of elected women in Scandinavian legislatures (Dahlerup Reference Dahlerup1988). The idea of critical mass was that after a certain proportion of women entered an organization a tipping point would be reached, after which change would occur. Dahlerup argued that growing numbers, while interesting, did not explain impact; instead it was necessary to consider what women legislators actually did. She concluded that it was critical acts rather than critical mass that explained change. Her argument could not have been more clear, it was pellucid. She never said that change was an automatic result of increasing numbers. But for the next two decades scholars cited her essay as an example of mistakenly using simple counts of women instead of their actions and attitudes to explain change (Dahlerup Reference Dahlerup2006).

While academic conventions do not require the misrepresentation of previous generations of scholarship, pressures to publish combined with preferences for ‘new’ discoveries undoubtedly contribute to the ‘fast science’ process that leads to it. There is a case therefore for supporting ‘slow science’ according to which close attention to earlier work is valued where it is relevant and new research is permitted to build on what has gone before without claiming to overturn it. Even in the rare event of a genuine paradigm shift (or scientific revolution à la Thomas Kuhn), the understanding of where we are going depends on having as accurate as possible an understanding of where we have been.

MODELLING GENDERED INSTITUTIONS

What does this preamble suggest about the supply and demand model of political recruitment? To recap, we developed our model from the standpoint of feminist political scientists who were looking to generalize from the processes we wanted to explore. We chose the model for pragmatic reasons and did not fully agree on how much it should frame our study of political recruitment. Certainly I never expected to be defending the model 25 years later and neither of us anticipated becoming the ‘straw women’ of the political recruitment literature in the way that we did.

Crucially, to answer our core research questions (who selects, who gets selected and how, does it matter?) we had to look at both selectors and selected. We observed and made use of the fact that that the selectors, party members, were themselves eligible, hence potential candidates who faced a series of stages in a selection process that in the UK was determined largely by party rules. The design involved matched surveys of selectors and aspirants, observations of selections and in-depth interviews with candidates, activists and officials and their mapping onto different stages in the selection process. Our surveys were linked to the British Election Study to facilitate subsequent elite/activist/voter comparisons. It was an intricate and time-consuming business that has not since been replicated on the same scale.

We were interested in the processes of recruitment in UK political parties in part because of frequent claims that women were absent from parliament (at the time of our initial study before the general election of 1987 only 3.7 per cent of MPs were women) because they did not come forward to be selected as candidates. Such commentary blamed the victims. We did not think it was true. Both of us frequently met women who did want to be MPs, who told stories of bitter struggles within parties and unions, of undemocratic and sexist blocking of their efforts. The issue was on the agenda because it was raised by women party activists, by the women’s liberation movement and the new and short-lived UK Social Democratic Party (Lovenduski and Norris Reference Lovenduski and Norris1993).

This was a political party issue. It is almost impossible to be elected to the British parliament as an independent; party endorsement is essential. In the 1980s there was a struggle inside the Labour Party over the issue. The other parties were not taking women’s claims for representation seriously at that time (Lovenduski Reference Lovenduski1993). Initially we attempted to examine the candidate selection process in order to determine where the blockages occurred and who was responsible. We quickly realized that in the UK the process itself was obscure, a matter of largely unregulated internal party rules, informal processes and hidden power relationships – the ‘secret garden’ of politics that Gallagher and Marsh (Reference Gallagher and Marsh1988) describe so vividly. No one had undertaken a systematic study for at least 25 years. After mapping out the problem (Lovenduski and Norris Reference Lovenduski and Norris1989) in terms of outcomes over time, we turned our attention to the selection process to try to determine at what stages different aspirants were excluded. Hence we first mapped processes and traced actors (from eligibility to election) in all the political parties, an undertaking not helped by the fact that British parties frequently change their selection procedures and sometimes change them dramatically after losing an election. We were dealing with considerable formal and informal institutional change.

Although the existing literature on candidate selection was out of date, there were some important insights that we found convincing. Previous studies by Ranney (Reference Ranney1965), Patterson (Reference Patterson1966) and Rush (Reference Rush1969) reported the importance of informal rules, accepted habits and practices, party cultures and local, national and regional contexts to outcomes. These findings and our acceptance of them pointed us towards the close examination of the selecting institutions in our research. Regrettably, but understandably, we did not think to locate ourselves in the emerging new institutionalist literature. This should not be surprising. The seminal works of March and Olsen (Reference March and Olsen1984) and Skocpol et al. (Reference Skocpol, Evans and Rueschemeyer1985) had only just been published, and the subsequent developments and explications of the new institutionalism industry were yet to come. So we were working on this very large project without the benefit of the vocabulary and concepts that today’s new institutionalism and feminist institutionalism provides. Even so, as I have argued above, I think our work fits comfortably into the approach.

FROM METAPHOR TO CONCEPT TO MODEL

Borrowed from economics, the idea of a supply and demand interaction seemed to fit our core question about whether selector demand or aspirant candidate supply explained the low numbers of women MPs. Vicky Randall (Reference Randall1987) and Jill Hills (Reference Hills1981) had both used the idea as a metaphor when writing about British political representation of women. We took it further, first to pinpoint the different stages of the recruitment process and then to assess if supply or demand explained who got through the final stages of the process. I now think we should have named the model more precisely as a ‘staged supply and demand model of political recruitment’ to emphasize the particular way we used the idea.

Initially we disagreed about this and to some extent that disagreement continued. At first I was doubtful and regarded the model to be an over-simplification that would lead to a data-driven account rather than a well-theorized one. Gradually as we discussed it and worked with it I realized that the model, for all its apparent simplicity, has considerable capacity to carry ideas, concepts and methodological refinements. It was and is basically up to the researcher how to operationalize it. This has been very apparent in research in the years since.Footnote ⁴ I am thinking especially about the work of Meryl Kenny (Reference Kenny2013) and Jeanette Ashe (Reference Ashe2014), who have developed the model in different ways. A further important advantage of the model was that it permitted and aided comparison across different party processes. Potentially it could also facilitate cross-national comparison, which is especially difficult in candidate research because both electoral and party system differences obscure the similarities in processes that are apparent in the stages that the model highlights. I discuss this point further below.

One of the most daunting challenges of the research was dealing with evidence from both quantitative and qualitative data, of deciding at various points how each set of data animated the model. I think the supply-side emphasis of the conclusions was too data driven and did not take our own extensive and rich qualitative material sufficiently into account. For example, there was plenty of qualitative evidence of a supply/demand feedback effect on the attitudes of aspirants and selectors but in the absence of explicit quantitative evidence we did not make enough of it.

This was not because we did not think about it, rather that we didn’t think we had enough evidence to generalize about the feedback and interaction of supply and demand. I mention this because it is an example of how the seductive ‘certainties’ of quantitative analysis tended to override other insights. To demonstrate the presence and impact of feedback effects in an analysis that satisfied the requirements of both kinds of data collection, it was necessary to repeat the study in successive elections and to gather data from party records about applicants for candidacy. We were never in a position to do this. We were, however, able to demonstrate the interaction of supply and demand dynamics at various stages of the process.

STAGES IN THE SUPPLY AND DEMAND MODEL OF POLITICAL RECRUITMENT

We agreed that an initial task was to assemble a huge amount of qualitative information about rules, processes and practices that would be organized thematically. Our development of the metaphor into a model took the form of the identification of recruitment stages in the form of a ladder of recruitment (Figure 1) that set out the stages from first application to election as a member of parliament. This ladder could be expanded or collapsed, extra rungs added for a more detailed examination of all or part of the process. For example, an additional distinction could be made at the top of the ladder between parliament and government or at the base between aspirants and applicants, or even between the ‘interested’ public and citizens. It is relatively straightforward to extend the ladder; the problem is collecting the relevant information – information that is frequently withheld from researchers by political parties (Figure 2).

Figure 1

Ladder of Recruitment

Figure 2

Extended Ladder of Recruitment

The utility of the ladder is that it permits the allocation of supply or demand factors to explain drop-off at each of its steps or stages. Because the selector survey was effectively a survey of party members, we were able to ask members if they had ambitions to become MPs. This permitted some insight into how the pool of applicants was constituted and also into the interaction between supply and demand. This aspect of the data set has not been much used by subsequent researchers, but it should be of use to those who are especially interested in ambition and motivation. Another advantage of the stages model is that it facilitates comparison. It permits the comparison of processes in different types of political parties and party systems, as this special issue clearly shows, including hybrid democracies such as those examined by Elin Bjarnegård and Pär Zetterberg (Reference Bjarnegård, Kenny and Zetterberg2016, in this issue). Even though the formal mechanisms of selection may vary, the stages are fairly similar, as are many informal practices and processes. Thus research on candidate selection in list systems could use the drop-off points in different parties to show which processes and stages most affect different types of would-be candidates.

The model can be used to show where and at what stage quotas should be applied if the numbers of different kinds of women and men are to be increased. Such research would contribute to more effective quota design. Indeed, the utility of the application of quotas at different stages is one of the problems addressed by all of the contributions in this special issue. We know that in the UK the lower down the pyramid a quota is applied, the less effective it is. As the contributions by Elin Bjarnegård and Meryl Kenny (Reference Bjarnegård and Kenny2016, in this issue) and by Niki Johnson (Reference Johnson2016, in this issue) demonstrate, localized processes – marked by informal practices of patronage and clientelism – are often used to subvert formal rule changes that clash with dominant interests. Even higher up we are learning that quotas have their limits, and that is why there are now feminist calls for legislative quotas that, in effect, bypass party preferences about candidate gender. Of course it is necessary to be careful about how such mechanisms are used. Tània Verge and Ana Espírito-Santo (Reference Verge and Espírito-Santo2016, in this issue) as well as Jennifer Piscopo (Reference Piscopo2016, in this issue) show that when legislative quotas do not fit with existing party practices and norms, they may only have limited effects on the numbers of women elected.

FEMINISM, POLITICAL RECRUITMENT AND AFTER

Feminism is first and foremost a way of thinking and hence might better be considered in the wider context of what social science offers to the understanding of ideas, especially how ideas are formed, developed, tested and advanced. The struggle to establish the legitimacy of feminist research is part of the politics of social science. This is a politics that is constructed around struggles over meaning. Steinmo (Reference Steinmo2008) goes a step further when he reminds us that the important battles in the social sciences are struggles over the meaning of ‘science’ defined in terms of its predictive capacity. Too often, he suggests, an obsession with prediction distorts research and obscures the importance of explanation. Steinmo’s observation has resonance for feminist research into politics. Feminists see gender relations as an instance of power relations. Feminist political scientists want to discover and explain gender effects in political life, a project that inevitably leads them to focus on how political institutions are formed and sustained and how gender is embedded in them.

Since the publication of Political Recruitment, a number of feminist and other scholars have explored aspects of gender and candidate selection at both national and subnational levels in the UK and elsewhere. The supply and demand model has come under considerable and rigorous scrutiny and has been utilized and developed in both quantitative and qualitative work (Ashe Reference Ashe2014; Kenny Reference Kenny2013). While most of the subsequent qualitative studies of UK candidate selection are gendered at least insofar as they take sex into account, only a few quantitative studies have appeared since 1995 and these do little more than note sex as a background variable (see, for example, the multinational study led first by Herman Schmidt and now by Wolfgang Müller).Footnote ⁵

In contrast, the British Candidate Study was a multi-method study in which the qualitative research explored aspects of gender and gender relations while the quantitative work relied on a binary sex variable, but took into account other aspects of difference (race, class, age) that might be incorporated into preliminary intersectional studies (see Jensenius Reference Jensenius2016, in this issue). Incorporating an adequate conception of gender, of masculinity and femininity into the quantitative work was then and is now a huge challenge. With the development of feminist institutionalism and its application to candidate research, we are better able to meet this challenge, as evidenced by the contributions in this special issue. I see no reason why, properly used, our staged version of the supply and demand model cannot both frame such analysis and advance the feminist contribution to political science.

So far I have identified two main challenges for feminist candidate selection research and hinted at a third. First is gendering the political recruitment research. For the purposes of this overview I treat gender as a scale of masculinity and femininity along which behaviour and attitudes may be ordered. Evidence of gender effects is found in the presence of codes, processes, norms and behaviour that reflect accepted (but possibly changing) dimensions of masculinity and femininity. Second is incorporating the concept of gender, properly understood, into the quantitative parts of recruitment research with a view to a fuller integration of quantitative and qualitative parts of research design. The third challenge is to construct well-designed comparative studies that can be a basis for theory building and for the identification of good and poor practice. The staged supply and demand model is likely to prove useful for meeting all three challenges.

Feminist candidate studies are to some extent mainly working on the first challenge. The supply side is still an issue and the absence of systematic selector research has distorted or at least inhibited the examination of feedback effects. Ultimately the gendered nature of political institutions is a matter for empirical research which tests, enables us to refine and so advances the theories we construct. Candidate research offers a good example of how important are the kinds of institutional contexts that pre-feminist researchers described. The rules of the game, the nature of the actors, the quality of the aspirants, the processes of decisions are all elements that we need to think about in terms of gender. While gender inequalities are largely unremarked in earlier accounts of political institutions, I am confident that it is easier to locate and explain them with the knowledge of how the institutions were seen to work by previous generations of scholars. It is not enough for us to know that parties may discriminate against women, we also need to know via what mechanisms and configurations the discrimination operates and what is at stake (see Kenny and Verge Reference Kenny and Verge2016, in this issue). The contributions to this special issue take up this challenge, providing a range of new data and insights into the ‘secret garden’ of candidate selection around the world.

Progress on the second challenge is more elusive. Yet, in any multi-method design, the insights derived from qualitative work such as in-depth interviews and focus groups are often used to frame the ideas and questions which will underpin ‘quantitative’ attitude and behaviour surveys. Qualitative methods are routinely used in multi-method designs to identify areas that it would be useful to operationalize quantitatively, usually in the form of survey questions. At the same time, results from quantitative studies frequently give rise to further qualitative exploration. Too often researchers forget that the data collection originated in ideas about what is important that in turn were developed via other research methods. This process is a given of many research designs but I wonder if it might not be more thoroughly examined by feminist scholars. Properly theorized survey questions will provide considerable analytical power, enabling the construction of scales of gendering that illuminate bias. So we might consider what kind of questions we need to ask if we want to make more insightful statements about gender roles in political recruitment. This is a big undertaking. What aspects of gender relations would we need to explore? Which aspects are amenable to large-scale survey research? What would the priorities be? How should we sequence the data we collect?Footnote ⁶ The staged supply and demand model offers a fruitful basis for designing and analysing survey questions that distinguish between supply-side and demand-side effects.

The third challenge, only hinted at here, is that of comparison and its role in theory building. This is work in progress, its theoretical utility well demonstrated by the articles collected in this special issue, each of which is comparative, gendered and contributes to the theorization of candidate selection. Such work will no doubt be built upon, further developed by these and other feminist scholars who can make good use of the staged supply and demand model. The model is sufficiently abstract that it should travel across national, system and party boundaries (assuming, non-trivially, that researchers can get the necessary access and information to identify, trace and quantify the effects of each stage). The conceptualization of political recruitment into discrete stages should be possible in research on candidate selection in any electoral or party system. Its utility is both scientific and practical.