Women continue to be underrepresented in politics. A contemporary focus on inequality, however, belies the remarkable changes that have occurred in women's political representation, participation, and impact over time. Thus, scholars are increasingly calling for longitudinal theory and research that focuses on change across time. In this essay, we advocate moving to a longitudinal approach, which, we argue, has four broad implications for theories on women and politics. First, in shifting from a static to a longitudinal perspective, we may find greater support for existing theories that have to date generated only weak or mixed empirical evidence. Second, in contrast, we may also identify limitations of documented findings, which may not generalize broadly across time. Third, some theories imply over-time processes but have never been modeled as such. Thus, a longitudinal approach promises new, and potentially more rigorous, hypothesis testing of existing theories. Finally, theorizing over time may generate new explanations for stasis, growth, or decline in women's political representation or participation that have not, to date, been considered.
Greater Support for Existing Theories
First, in moving to a longitudinal approach, we may find greater evidence to support theories or perspectives that have thus far generated null or mixed results. That is, in some cases, testing our theories in longitudinal models might show more effects than in cross-sectional research designs. To provide an example, it is possible that once we account for time, supply-side theories of women's access to political power will find greater empirical support than in existing research.
In women in politics research, supply-side arguments continue to maintain a prominent position in the list of reasons that women are underrepresented in political office (e.g., Dolan, Deckman, and Swers Reference Dolan, Deckman and Swers2007; Paxton and Hughes Reference Paxton and Hughes2007). Supply-side theories hold that women need human and financial capital, gained through educational and work experience, to stand for office. In the American context, therefore, in states where women make up a greater proportion of law students or of the labor force, they are also expected to fare better politically (Arceneaux Reference Arceneaux2001; Norrander and Wilcox Reference Norrander, Wilcox, Thomas and Wilcox2005; Oxley and Fox Reference Oxley and Fox2004; Sanbonmatsu Reference Sanbonmatsu2002).
When we look across countries, however, empirical evidence to support this perspective has been weak or mixed (e.g., Gray, Kittilson, and Sandholtz Reference Gray, Kittilson and Sandholtz2006; Kenworthy and Malami Reference Kenworthy and Malami1999; Kunovich and Paxton Reference Kunovich and Paxton2005; Paxton Reference Paxton1997). Despite inconsistent support for supply-side theories in the international context, it is difficult to dismiss the idea that women's education or workforce experience is politically relevant. It simply makes intuitive sense that a greater supply of educated, working women should produce higher numbers of women in politics.
Adopting a longitudinal approach may help to reconcile the disconnect between theory and evidence. In a number of Western countries, women made their gains in education, the labor force, and politics decades ago. By focusing only on recent time points, research may be masking substantial historical variation in women's social structural position that could have important explanatory power for women in politics today. Further, rather than comparing countries with high levels of educated women to countries with low levels, it may be more important to focus on over-time change in women's status within countries or states (e.g., Norris and Lovenduski Reference Norris and Lovenduski1995). If we employ longitudinal models, we are able both to focus on change within countries and to include the period of time when women were making gains.
Limits to the Generalizability of Existing Findings
In addition to finding greater evidence for theories that have to date received little empirical support, we can also expect to encounter the opposite—some of our research findings may not generalize well across time. As a speculative example, consider arguments that political parties on the left end of the political spectrum are more likely to promote traditionally underrepresented groups such as women (Caul Reference Caul1999; Matland Reference Matland1993). Cross-national research on women's political representation in recent cross sections has consistently documented that countries with a dominant leftist party have more women in political office (Kenworthy and Malami Reference Kenworthy and Malami1999; Reynolds Reference Reynolds1999; Rule Reference Rule1987). We might, however, speculate that when we look more broadly across history, we may find that leftist political parties have not always advanced women to a greater degree than have right-wing parties.
There are reasons to believe that leftist parties may not always have been better for women. In the United States, for example, research suggests that prior to the 1980s, the Republican Party was just as likely to advance women politically as was the Democratic Party (Dolan, Deckman, and Swers Reference Dolan, Deckman and Swers2007; Paxton and Hughes Reference Paxton and Hughes2007). Indeed, the first woman elected to national legislative office in the United States, Jeannette Rankin, was a Republican. Although today female Democrats outnumber their Republican counterparts at both the national and state levels, this was not always the case. In the House of Representatives before 1965, Democrats and Republicans traded prominence in women's representation (CAWP 2001).Footnote 1
Another reason why leftist parties may not always have advanced more women is that historically, women in many countries have held more conservative ideological positions, on average, than their male counterparts (Inglehart and Norris Reference Inglehart and Norris2000; Klausen Reference Klausen, Klausen and Maier2001). Indeed, in 1960, Seymour Martin Lipset found that in “practically every country for which we have data … women tend to support the conservative parties more than do men” (p. 221). And although today there is evidence that across most advanced industrial societies women have moved to the left of men (Inglehart and Norris Reference Inglehart and Norris2000), women's historical conservatism may mean that in the past, leftist parties were not necessarily more beneficial for women.
To be clear, understanding women's representation, participation, and impact in the contemporary period is critical. Certainly, a contemporary understanding of party ideology and gender dynamics helps scholars advise policymakers and party leaders. But a longitudinal approach is also necessary for a full understanding of women's experience throughout history. Thus, we need to investigate whether theories developed and tested only in recent time periods apply equally to the past. More generally, a longitudinal approach reminds us that broad historical shifts are possible and that we cannot project too far into the future from current trends.
Better Tests of Existing Theories
It is also important to recognize that some of our theories are inherently longitudinal. That is, some theories rest on notions of change or progression more than others. Such theories are likely best tested with a longitudinal perspective. For example, consider growing interest in the impact of gender quota laws on women's political representation (e.g., Baldez Reference Baldez2004; Dahlerup Reference Dahlerup2006; Krook Reference Krook2007). The introduction of a quota—a legislative or party rule that requires a certain percentage of candidates or legislators to be women—is essentially a temporally bounded change in election procedures. It therefore follows that an ideal way to determine whether quotas impact women's political representation is by allowing for over-time change. Through cross-sectional designs, we already know that countries with quotas have more women in politics than countries without quotas (Tripp and Kang forthcoming). There is an alternative longitudinal way to ask the question: Do numbers of women in politics rise from previous levels when quotas of particular types are introduced? Answering that question requires longitudinal data (both previous and current levels of women's parliamentary representation) and longitudinal methods (for example, an interrupted time series).
A longitudinal approach could also enhance our tests of the impact of women on public policy or legislative style. Realistically, it is difficult to separate the impact of a female legislator on public policy from her political party or constituents (Dolan, Deckman, and Swers 2006: chap. 7; Swers Reference Swers and Rosenthal2002). In a nutshell, is it the fact that Nancy Pelosi is a woman that makes her vote a particular way, or the fact that she is a Democrat or that she represents a very liberal constituency? Some research on women's impact controls for constituent and party effects (e.g., Bratton and Haynie Reference Bratton and Haynie1999; Schwindt-Bayer Reference Schwindt-Bayer2006; Swers Reference Swers1998). But a longitudinal approach could help us gain additional leverage on the question by accounting for change over time—focusing on constituencies that switched from men to women or women to men. Since most constituencies are reliably Democratic or Republican, looking across time within constituencies would be a strong control for alternative explanations of women's roll-call votes or bill introduction (see, e.g., Gerrity, Osborn, and Mendez Reference Gerrity, Osborn and Mendez2007). Similarly, over-time designs could also help us determine whether, in terms of legislative style, women change institutions or institutions change women (Lovenduski Reference Lovenduski, Lovenduski and Norris1993, 6).
Longitudinal research designs can be difficult to construct, however, and of necessity, they introduce complexity. To continue the example of women's impact, in implementing a longitudinal design, a researcher might need to account for changes over time in the boundaries of congressional districts, the changing composition of congressional districts, and other factors that would undermine the direct comparison of a male legislator and a female legislator. Further, the small number of women who have attained congressional seats is an obvious limitation to this research design. Our point is certainly not to suggest a perfect longitudinal design here. Instead, we simply hope to point out the potential benefits to thinking creatively about incorporating time into our tests of hypotheses.
In general, a longitudinal approach to testing existing theories is likely to be most useful for theories that center around change, development, or progression. Further, longitudinal methods are well suited for addressing questions of causality and endogeneity where they exist in our theories (e.g., cultural attitudes and women's political representation are likely in a reciprocal relationship). Moving in a longitudinal direction will allow researchers to explain change and to precisely demonstrate cause.
New, Explicitly Longitudinal, Theories
A final implication of a move from static to longitudinal thinking in women and politics research is that new, explicitly longitudinal theories may be developed. A longitudinal perspective suggests forces for change that may vary across time but not across countries, states, or individuals. As an example, consider how the international women's movement grew in size and power over time (Berkovitch Reference Berkovitch1999; D'Itri Reference D'Itri1999; Paxton, Hughes, and Green Reference Paxton, Hughes and Green2006; Rupp and Taylor Reference Rupp and Taylor1999). From just a few organizations in Western nations in the late 1800s, the international women's movement ultimately grew to encompass more than 40,000 women and men from over 180 countries who came together in Beijing for the Fourth Global Conference on Women. Further, the movement increasingly cooperated with international agents, such as the United Nations, that diffuse global norms to nation-states (Meyer et al. Reference Meyer, Boli, Thomas and Ramirez1997). Thus, through both global expansion and collaboration, the international women's movement was more and more able to transmit a discourse of substantial gender inclusion to nation-states over time. We would therefore expect the impact of the international women's movement on women in politics to vary over time, but to be generally uniform in its effect across countries.Footnote 2
Alternatively, consider period effects, which can also be assumed to vary across time but not across units. Continuing with the international women's movement example, one could argue that the messages and goals of the international women's movement regarding women's political incorporation have changed and evolved over time (D'Itri Reference D'Itri1999; Paxton, Hughes, and Green Reference Paxton, Hughes and Green2006). In the early years of the movement, pressure on states centered on the basic provision of political citizenship to women (the right to vote) (Rupp and Taylor Reference Rupp and Taylor1999). The language adopted by international institutions during this period therefore focused on “political rights” for women. Later, a second wave of the women's movement emerged that contested women's status on a broader scale. In the realm of politics, the international women's movement began to call for policies to increase women's representation. The discourse of the second wave thus shifted from “political rights” to “women in political decision making.” Beginning in the 1990s, the discourse changed once more to emphasize specific thresholds or targets for women in political decision-making positions. The discourse continued to concern representation, but now phrases such as “critical mass,” “gender quotas,” and even “gender balance” were stressed by women's groups. This changing discourse suggests that pressure from the international women's movement for women's incorporation progressed in identifiable periods over time. Within each time period, therefore, pressure to respond to these changing messages would be uniform across states.
An alternative theory suggested by taking a longitudinal approach is that the fate of women may be tied to the fate of political parties. In brief, political parties gain and lose power over time. Political parties also differ in the extent to which they promote women as candidates. It follows that some of the observed fluctuation in women's representation over time (e.g., Dominica moved from 17% women in 1990 to 9% women in 1995 and then back to 19% in 2000) could be due to shifts in power across political parties (e.g., Dominica's three political parties were gaining and losing seats across those elections) (IPU 2007). The theory we propose is slightly different from existing theories of the importance of political parties for women. Existing theories focus on party rules (Caul Reference Caul2001), party contagion (Matland and Studlar Reference Matland and Studlar1996), party dominance (Matland and Brown Reference Matland and Brown1992; Rule Reference Rule and Whitaker1999; Sanbonmatsu Reference Sanbonmatsu2002), or elite promotion of female candidates (Caul Reference Caul1999; Kunovich and Paxton Reference Kunovich and Paxton2005; Sanbonmatsu Reference Sanbonmatsu2006). In contrast, the longitudinal theory we are suggesting is focused on party success and the rise and fall of political parties over time. Obviously, these are not the only new theories that might arise from taking a longitudinal perspective. Our purpose here is simply to show that new explanations are possible when we begin to think about time and history.
To summarize, in this essay we have attempted to illustrate four implications of moving from static to longitudinal research on women in politics. On the one hand, we suggested that for theories with only weak support in contemporary cross-sections, we may find greater support when looking into the past or within countries over time. On the other hand, we also suggested that some of our documented findings may be a product of current dynamics and may not generalize broadly across time. We also stressed that existing theories may be implicitly longitudinal and therefore better tested with longitudinal data and methods. Finally, we suggested that new theories may arise from increased attention to time and history.
The implications we have discussed here are likely not the only consequences of moving to a longitudinal approach in politics and gender research. Indeed, incorporating time may make an important contribution to our knowledge in ways we cannot anticipate from our current, largely cross-sectional perspective. We could go so far as to suggest that a new focus on time and history could revolutionize our understanding of women and politics. Toward that end, we advocate that researchers seriously consider the introduction of longitudinal theories and methodologies to their work.