|Anne Marie Goetz
Background Paper for the Education For All Report, UNESCO, 2004
Women’s Education and Political Participation
It seems evident that formal education should be strongly associated with political participation for women and for men. Indeed, the American sociologists Burns, Schlozman and Verba assert, on the basis of decades of research into the factors influencing women and men’s engagement with politics in the USA, that education is an ‘especially powerful predictor of political participation’ (2001:286). They identify a range of direct and indirect effects that formal education has upon political participation. Its direct effects include the acquisition of the knowledge and communication skills useful for public debate, and direct training in political analysis through courses with current events content. Its indirect effects are many and include the benefits of voluntary engagement in school government, clubs, sports, and school newspapers; these arenas provide young people with an early apprenticeship for politics, where they can exercise leadership, develop civic skills of cooperation and negotiation, and acquire bureaucratic and organizational skills useful for political activity. Education enhances other factors supporting political engagement, such as access to high-income jobs that provide the resources and contacts for political activity, and access to non-political associations such as charitable organizations or religious establishments that can be a recruitment ground for political activity (ibid: 141-2).1
There is a wide variation between countries, however, in the relationship between women’s education levels and their representation in formal politics, and their participation in other political activity. The United States, which outranks other industrialized democracies in terms of the numbers of women in higher education (and in the work force, and in professional positions), has seen persistently low numbers of women in formal politics, reaching an all-time high of just 14.3% of Congress in 2002 (Center for Voting and Democracy, 2003). Uganda, Rwanda, and Mozambique, among the poorest countries in the world with female adult literacy levels of just 41, 60.2 and 28.7 percent respectively, have parliaments in which between 25 to 30 percent of legislators are women. This contrast suggests that the connection between education and engagement in formal representative politics is not directly observable, and invites us to explore the nature of the relationship between women’s education and political participation. This paper begins by distinguishing between political representation and political participation more broadly. It assesses the importance of education among the many factors commonly associated with individual and group political engagement and effectiveness. This is done by reviewing recent statistical analyses of the relationship between these factors and variations in women’s participation in formal politics around the world. The paper ends with a consideration of the role of women’s education in advancing their interests at the level of local government, which has seen rising numbers of women participants in countries around the world.
Political participation matters a great deal for women as a group and as individuals. Whether women work together to protest gender-based injustices or whether they participate in non-gender-specific associations and struggles, the most important group benefit from political participation is influence on decision-making to make public policies sensitive to the needs of the group in question. For groups, participation also builds social trust and capital, and provides a form of democratic apprenticeship; it offers socialization in the norms of reciprocity and cooperation, the capacity to gain broader perspectives on particular problems in order to develop a sense of the common good. For individuals, political participation builds civic skills, while successful lobbying can result in improvements in personal welfare and status. Explanations for the very slow progress women have made in gaining political office around the world have been multi-causal, including: their lack of time for politics due to their domestic obligations, their lack of socialization for politics, their lower social capital and weaker asset base than men owing to discrimination in schools and in the market, their underrepresentaiton in the jobs that favor political careers, their marginalization within male-dominated parties, their inability to overcome male and incumbent bias in certain types of electoral systems (Randall, 1987; Matland and Taylor, 1997; Rule, 1981).
Women’s political participation is most often measured in terms of the numbers of women to be found in formal politics, in positions of public office to which they have been elected. This extremely crude measure is made even more so by the tendency to limit it to the numbers of women in the main legislative house at the national level, excluding not just numbers of women in regional and local government, but numbers of women elected as magistrates, members of the boards of public bodies such as schools or health facilities, and the like. The reasons for using this measure have to do with simple convenience. There are significant data gaps on numbers of women in local governments and other sub-national elected bodies around the world, and there is such wide variation in governance systems for sub-national communities and public bodies that they are barely comparable.
Numbers of women in representative politics are not the best indicator of the extent and intensity of women’s political participation because there is no necessary relationship between the two. Relatively large numbers of women were found in politics in socialist countries in periods when women’s independent civil society activity was suppressed under single-party governments (Molyneux, 1994). Relatively large numbers of women are found in local governments in some countries in spite of the fact that the women’s movement can be weak at these levels – for instance in France or Uganda. And India and the USA, with the largest women’s movements in the world (in terms of the sheer number and variety of women’s organizations) have some of the lowest levels of women in national office.2 Numbers of women active in women’s organizations, or at least numbers of women’s organizations in a country, might be a better indicator of levels of women’s political participation. Data on the strength of the women’s movement in a number of countries has been compiled by Ramirez, Soysal, and Shanahan (1997). Their indicator shows the number of women’s organizations at the national level in 1990 whose mission is expressed by themes of emancipation, political participation, democracy, or socialism. However, organizations of this type may be urban-biased and elitist and unrepresentative of the interests of the majority of women, and the existence of a large number of such organizations in a country may indicate not strength but fragmentation and therefore weakness in the women’s movement.3 Nevertheless it is one of the few measures available of women’s activism. This variable has not, to my knowledge, been correlated with different measures of women’s educational achievement (adult literacy, numbers of women's college and secondary school graduates, enrollment ratios). It would be a relatively simple statistical exercise to do this and worthwhile for the purpose of this EFA report.
Women’s political participation is best understood more broadly than numbers of women in office, and indeed, more broadly than numbers of women’s organizations, as women may express their political interests through participation in a wide variety of political and civic associations. Verba, Schlozman and Brady define political participation as ‘activity that has the intent or effect of influencing public action, either directly, by influencing the making of public policy, or indirectly, by influencing the selection of political decision makers’ (1995:38). This definition includes voting, campaigning for a party or supporting party work through other means (e.g. policy development, membership drives), contacting policy-makers directly by writing or telephone, protest activities, getting involved in organizations that take a stand in politics, taking part in informal efforts to solve community problems, and serving in a voluntary capacity on local governing boards such as school or zoning boards. This definition is clearly culturally-specific; notions of citizen lobbying of representatives or participating in political campaigns apply best in democratic contexts that lack violence and corruption in political competition (particularly in electoral campaigns), and that have disciplined parties with internal democracy, clear programmes and positions.4 This narrow definition has also been criticized by feminist political scientists as being overly focused on individual political acts and for excluding the forms of public engagement favored by women. Most insist that women’s civil society activity – what Verba et al would call non-political activity – be included in the definition of political participation (Molyneux, 2001; Marshall, 2002; Tripp 2000). Others argue that the definition of ‘political acts’ must include resistance to injustice in the private sphere, for instance in the family or the firm.
The difficulty with overly broad definitions of political participation is that activities are hard to measure, particularly acts of resistance in the private sphere, and cross-nationally comparable data are simply not available. Burns et al developed an eight-point scale of political participation that measures acts that range from voting to protest, voluntary activities in parties, religious associations, and communities (2001). This has been useful for describing differences in levels of political engagement between women and men in the USA, and for explaining its causes, but has not been tested elsewhere. They find that while women are as or more likely as men to vote, they are significantly less likely to engage in a range of other political acts, such as contacting their representatives, contributing in cash or time to campaigns, or joining a political organization (2001:2).
It is worth paying some attention to the findings of the Burns et al late 20-century survey of factors influencing the political activity of 15,000 American adults as this is one of the most sensitive studies available. They tested the influence of the following factors on the propensity of women and men to participate in political activity: a childhood socialization in politics, educational levels, participation in high-school clubs, employment in jobs providing political connections and opportunities (e.g. professions, or unionized work), participation in non-political organizations, participation in religious organizations, available time, family income, own income, and experiences of gender-based discrimination. They found that men’s advantage in political participation was linked to a much stronger endowment of two key factors: education, and the types of jobs that provide the resources and contacts needed for politics (252). Contrary to popular expectations that time constraints and a lack of resources inhibit women’s political engagement, they found that leisure time did not differ between women and men, and that it was not related to political participation, and that while family income did have a significant impact on political activity, levels of family income differed little between women and men.th Women were found to have higher endowments than men of some factors positively related to participation: participation in high school clubs and in religious associations. Interestingly, women’s experience of gender-based discrimination also produced political activity, though the study failed to measure how far this discrimination simultaneously eroded other participatory resources (259). However, women’s endowments of these participatory factors were outstripped by men’s educational and employment advantages. Also, women’s religious affiliations have ambiguous implications for their subsequent recruitment into political activity, given that some religious institutions keep women out of leadership positions.
The key finding in Burns et al is that ‘gender differences in participation are the result of disparities in the stockpile of factors that facilitate participation, not of gender differences in the way participatory factors are converted into activity’ (2001:259). This reinforces what, according to the political philosopher Anne Phillips, ‘everyone knows to be the case: that the extent to which individuals become involved in politics and thereby gain access to decision-making channels is directly correlated with the resources they have at their command; that all else being equal, those who have everything else get political power as well’ (1991:79).
This common-sense explanation of men’s advantage in political engagement goes a long way in explaining low levels of women’s political participation. However, Burns et al do not consider women’s participation in formal political institutions – in representative politics. Their study does not, therefore, help to explain why it is that even when women’s educational levels approach parity with those of men, formal political institutions remain relatively closed to women. Simply put: women’s participation in formal politics does not appear to increase in step with advances in their educational status in comparison with men. Women’s educational attainments in the USA now equal those of men, yet the persistently low numbers of women in representative positions – below the already low global average – suggest that there may be something specific to political institutions that discourages female participation. The observation about a lack of relationship between women’s education and their achievements in formal politics holds for other countries too: Jayaweera’s study of 23 middle and low income countries in Asia shows little significant difference in the level so of women in formal politics regardless of whether there is near universal education, such as in the Republic of Korea, whether there is quite extensive female educational participation (Sri Lanka), or whether there is extremely low female literacy (Pakistan and Nepal) (1997: 421). In other words, the absence of a strong linear relationship between women’s educational attainments and their numbers in formal politics suggests that there must be something specific to political institutions that discourages female participation. Jayaweera suggests that it is no great mystery: women’s many time constraints from their domestic and other work commitments, and ‘gendered perceptions of political and community leadership’ mean that women both elect to stay out of politics, and are not considered admissible as representatives even if they do wish to participate. We shall return to this problem shortly. Thus the suggestion in Burns et al that women can, as easily as men, convert endowments in ‘participatory factors’ into participatory activity, is contradicted by qualitative research into the persistence of gender-based selection and treatment biases in important non-political and political institutions, even in as deeply established a democracy as the USA (Kenworthy and Malami, 1999; Keiser et al, 2002).
In assessing the findings in Burns et al, allowance must be made for the fact that gender disparities in education are greater in older age groups in the USA – those above 40 and most likely to hold formal representative positions – than in the generation under 30 currently enjoying parity in educational achievements. Therefore we may anticipate that women’s political participation will increase as the better-educated younger generation of women mature. Indeed, because the Burns et al sample includes a substantial proportion of over-40s (mirroring their proportion of the population), it overstates the current gender gap in educational achievements, but does accurately reflect the way a residual gendered educational gap affects the political participation of the over-40s. Whether a measurable time delay exists between increases in women’s educational status and their participation in formal politics is nor clear from existing studies, and this would be a worthwhile question for the EFA team to test, using available statistics.
While the general pattern of gender-differences in participation observed by Burns et al may well hold for many other nations, there are no cross-national studies of gender-based variations in the kinds of political activities they measure, mainly because of a lack of consistent data on gender differences in voting behavior, protest activity, voluntary community activity and so on. Explorations and explanations of gender gaps in political activity in other cultures must be sensitive to differing opportunities available for political participation given variations in political institutions and cultures.
Given the difficulties of measuring the quantity and nature of women’s political participation cross-nationally, we fall back upon the number of women in office, currently the only consistent and comparable source of data showing variations in women’s engagement in politics. Though far from an ideal indicator of levels of women’s political engagement, it is not entirely unrelated to the question of women’s relative political effectiveness in any particular country. The presence of more than average (currently the global average is about 15% of lower houses - IPU 2003) numbers of women in politics should indicate that some of the many obstacles to women’s political participation have been overcome. Overcoming any of these obstacles is to some extent contingent upon the success of the women’s movement or other civil and political associations in challenging the biases that differently select women and men into social, economic, and political institutions, and produce unequal and unjust treatment of women once they do gain access. Therefore the number of women in office must at least in part reflect the strength and achievements of women’s political activism. Attentiveness to the numbers of women in office is also not irrelevant to the project of ensuring that participation in the public arena to advance women’s interests. Though women in office are almost always social and political elites lacking connections to the women’s movement there is evidence from around the world that women legislators, even when in an acute minority, help to steer political debate in parties and legislatures to issues of significance to women and children (Lijphart, 1991; Rule and Hill, 1996; McDonagh, 2002; Thomas, 1994; Vega and Firestone, 1995; Chattopadhyay and Duflo, 2001).
Gender and Political Office: Determinants of Selection
A number of cross-national comparative studies of the determinants of variations in numbers of women in office have been produced since the early 1980s. All of these seek to establish the relative contribution of political factors (the design of the electoral system, the partisan composition of government), socio-economic factors (women’s educational levels, labor market strength), demographic factors (population size, levels of urbanization) and cultural factors (dominant religion) in determining the proportion of legislative seats held by women. Data on these variables are most consistent for a sample of 15 to 25 most developed democracies, and analyses of these have found political factors, particularly the distinction between Proportional Representation (PR) and majority-plurality electoral systems to be the strongest predictor of women’s electoral success, followed by educational attainments, labor force participation, and cultural factors, such as the divide between Catholicism and Protestantism5 (Norris 1985, 1987; Darcy, Welch, and Clark, 1994; Lovenduski and Norris 1993). Leaving out the design of electoral institutions, Rule found that 70% of the variation in the numbers of women in politics in 19 OECD democracies between 1987 and 1991 was explained by a combination of the number of women college graduates, women in the workforce, the length of women’s suffrage, and the level of unemployment (Rule 1994:20).
Applied to a broader set of countries that includes developing nations, these types of studies have had less consistent findings. Paxton’s analysis of 108 countries found effects for political and cultural factors but not for socioeconomic variables. Two other studies are more ambiguous, finding stronger results for socio-economic and cultural factors than for political variables (Oakes and Almquist, 1993; Matland 1998). Matland, for instance, compared a sample of advanced industrialized democracies with a sample of developing country democracies. He found that whereas a shift from a majoritarian to a PR system in an industrialized democracy would produce a 15.6% jump in the female proportion of the national legislature, this electoral system variable had a coefficient ten times smaller for developing countries (1998:115 – 117). The percentage of adult women with a university education in industrialized democracies had no statistically significant effect on the numbers of women in legislatures. Data on women’s higher educational attainment was not used for his developing country sample, and instead a composite measure of women’s cultural standing in relation to men was used. This brought together three measures: the ratio of women’s to men’s literacy, the ratio of women’s labor force participation to men’s, and the ratio of university educated women to men educated in universities. Matland found, in a regression analysis of 16 less developed countries, that this measure of women’s comparative standing in relation to men explained more variance in women’s share of legislative seats than did the electoral system variable (women’s comparative cultural status explained 2.5% of variations in numbers of legislative seats, while electoral systems explained just 1.56%). However, he found that this finding was not robust. The cultural variable had been used in tandem with a measure of development, and he found that the two had split the variances between them with all the positive factors loading on the cultural variable and the negative factors onto the development variable. When he dropped the development variable, the cultural variable plummeted to less than half its previous size and was no longer significant (ibid: 118). He had to conclude that women’s representation in LDCs ‘may largely be determined by idiosyncratic conditions within that country and not by broad forces influencing all LDCs’ (ibid: 118). Oakes and Almquist’s regression analysis for 93 countries found women’s labor force participation to be significant, accounting for the greatest amount of variation in women’s legislative positions (1993:76). Interestingly, they found no positive effect for the proportion of women in managerial positions, arguing that ‘women’s skills and experiences in positions of one sector of authority are not necessarily [felt] in other sectors’ (ibid 78).
These studies were conducted on data from the 1980s, and did not capture the effect of the global wave of democratization in the late 1980s and early 1990s. Two studies that use data from the 1990s are Reynolds’ comparison of factors determining the access of African women to legislatures with factors explaining numbers of women in politics globally (in a sample of 180 countries) (1999), and Kenworthy and Malami’s study of 146 countries (1999). These studies broadly distinguish between political variables, socio-economic variables, and cultural variables in clustering factors held to determine proportions of women in politics. Though they differ in the variations they chose to highlight between political systems, and in the proxies used to measure culture, both find electoral systems to be the strongest predictor of women’s political recruitment, with variants of Proportional Representation systems producing the highest numbers of women in office. Other variables found to be significant include the length of time since the extension of women’s suffrage (and hence public familiarity with the idea of women having access to political office), higher levels of political competition (the number of parties), and the presence of strong left-of-centre parties. They find religion, as a proxy for culture, to have a significant effect on women’s representation.
Results for socioeconomic variables are much weaker. Women’s educational attainment (female share of enrollees in secondary education6), female labor force participation, the number of national women’s political organizations, and national economic wealth were not statistically significant in explaining women’s share of seats in the legislature. One study, however, drew out the significance of women’s share of professional occupations and found this to have a statistically significant effect on women’s electoral success, but an effect distinctly secondary to the political variables of the electoral system and the left orientation of government (for 116 countries, the absolute t-value for the electoral system was 3.28, for left-wing government was 4.62, and for the number of women in the professions, it was 2.42) (Kenworthy and Malami, 1999:251). As they note: ‘It is not paid work per se that affects women’s opportunities for legislative representation, but rather the particular types of paid work in which women participate’ (Kenworthy and Malami, ibid).
The proportion of women in the professions is of course related to female educational attainments and so the influence of education on women’s electoral prospects might have been better captured by a measure of women’s tertiary-level enrollment or college degree-holding. Interestingly, however, Kenworthy and Malami found that when they substituted both university-level education measures, and a variable representing the share of women aged 25 – 64 with a university degree, for their measure of the female share of secondary education enrollees, their results did not change (ibid, 264 fn 13).7 They do not explain this outcome, but it may reflect variations in employment opportunities between nations, where class and cultural factors can keep even well-educated women out of professional occupations. This is suggested in Jayaweera’s study of female education and employment in Sri Lanka, where secondary and higher education does not produce occupational mobility for rural and working-class girls (1997:418). Kenworthy and Malami’s finding regarding the importance of women’s share of professional jobs contrasts with the finding in Oakes and Almquist’s earlier study bout the lack of significance of the variable measuring the female share of managerial positions. These contradictions are either the result of measurement errors and variations, or the result of changes over the past decade in the capacity of women in managerial positions to move into politics.
The finding as to the key role played by electoral systems is not surprising. Electoral systems can be understood as the rules determining how votes should be assigned to seats, and these rules have been shown to favor elites and men in majority-plurality systems. Single-member constituencies with first-past-the-post systems for determining the winner create incentives for parties to front the one candidate likely to appeal to a sweeping majority in a voting district, a ‘lowest common denominator’ (Reynolds 1999: 8) candidate who will not be a woman or a class, ethnic, or racial outsider (see also Norris and Lovenduski 1995; Inter-Parliamentary Union 1997; Rule and Zimmerman, 1992; Matland, 1999). But if more than one person can represent a constituency, and seats are assigned according to the percentage of votes won, then all parties have an incentive to embrace greater diversity, and so the resistance to fronting women candidates drops. The implications of this for women’s electoral prospects are best demonstrated in countries that use a combination of the two systems. For instance in Australia and Germany, different electoral systems are used for different levels of government. In both countries, fully three times more women are elected in mulit-seat districts using proportional representation than in single-seat ‘winner takes all’ competitions (Rule and Hill, 1996). In South Africa’s local government elections of 1995/6, a PR system was used for 40% of seats on local councils, and a ward system for the remaining 60%. For every party participating in those elections except the right-wing Freedom Front and the left-wing PAC (from which, respectively, just nine and two women were elected), the PR system returned far more women to local government office than did the ward system. In the ANC, of the 911 women elected (24.6 per cent of the total ANC winners), 717 won through the PR system and only 194 through the ward system (IDASA, 1997).
In a review of 53 national legislatures in 1999, national assemblies in PR systems were composed of on average nearly 20% women, compared to nearly 11% in majoritarian systems (Norris 2000). If we consider African countries on their own, countries with PR systems had legislatures in the mid-1990s in which on average 11.65% of representatives were women, in contrast with an average of 5.46% in countries with majority-plurality systems (Yoon, 2001:181). Though PR clearly returns more women to office, it works best in combination with efforts by parties to support women candidates, and in contexts favorable to notions of gender equity. Although almost all of the few countries in where women are one quarter or more of the national assembly have PR systems, they also share at least one of the following characteristics: an egalitarian political culture (the Scandinavian countries), strong socially egalitarian or left-of-centre dominant parties that have voluntarily adopted quotas of women on their electoral lists (Scandinavia again, and South Africa, Namibia, Mozambique), or, in a few cases, laws to institutionalize female quotas on party lists (Argentina, France).8 In Africa, only those countries with party quotas for women candidates rise above the 11.65% average proportion of women in office (Mozambique, with a 30% female parliament, and South Africa, with a 29.8% female parliament) (Yoon 2001: 181).
Other variations in PR electoral systems are significant for women’s chances of gaining office. District magnitude has a positive effect (the more representatives per district, the easier it is to overcome voter reluctance to select women candidates) (Rule and Shugart, 1999). Closed-list systems in which voters select the party but do not make their own selection among candidates seem to have some advantages in delivering larger numbers of women to office (provided the party supports women’s candidacies), and they are also seen to offer a form of protection to women who wish to campaign on electorally risky gender-equity policy platforms appealing to a cross-section of the electorate. But there are concerns about democratic deficits in closed-list systems, for instance the way they detach representatives from the citizens they are supposed to represent, making them accountable not so much to voters, but to party bosses.
Some first-past-the-post systems have attempted to compensate for the gender biases in the electoral system by reserving seats for women. Tanzania, for instance, reserves 17.5% of Parliamentary seats for women. Uganda reserves one seat in each of its Districts for women-only competition, in effect reserving about 20% of Parliamentary seats for women. Reserved constituencies have been explored at local government levels, with the most striking cases being the reservation of 30% of local government seats for women in both Uganda and India. Reservations of seats for women-only competition, or quotas for women candidates in political parties, are seen as non-democratic tampering with political institutions and therefore as necessarily temporary measures.9
As two long-time observers of women in politics note, PR systems with high district magnitudes generally can help overcome social barriers to women’s successful recruitment to Parliament, but the opposite does not hold: ‘favorable societal conditions cannot make up for unfavorable electoral arrangements’ (Rule and Shugart, 1995). This has implications for policy aimed at enhancing women’s capacity to participate in public decision-making. A long-held assumption in liberal feminist arenas is that women will be more successful in politics once gender inequalities in education and employment are reduced. As Kenworthy and Malami note, even though their analysis shows that women’s movement into professional occupations does advance their opportunities in the political sphere, that is the only one of five socioeconomic factors they consider that matters, and its effect is weaker than those of political and cultural determinants (1999:260). They conclude: ‘Altering political institutions and cultural beliefs and practices would seem to be a more effective, albeit not necessarily and easier, route to gender political equality’ (ibid).