A well-known Achilles heel of multiculturalism is the question of its effect on women. Rajeswari Sunder Rajan, for instance, discusses the top-down role of the state in India in “supporting … discrimination against women” by recognizing the primacy of religious community identity in personal law” (Reference Rajan2003, 16). Susan Okin's Reference Okin, Cohen, Howard and Nussbaum1999 essay “Is Multiculturalism Bad for Women?” perhaps engages this problem most directly. Okin avoids giving a direct answer to her question, arguing in a later essay that “those in the best position to answer it, in each specific context, are the women who are at the intersection of the issue—those within whatever minority cultural or religious groups are claiming group rights as necessary to preserve their group values and ways of life” (Reference Okin and Spinner-Halev2005, 72). This formulation, however, suggests a second question, to be addressed extensively in this article: Are patriarchal practices, gender inequality, or other limits of liberal rights permissible if they have the approval of the women who are members of the group?
From a normative perspective, this article questions multiculturalists' prioritization of the collectivity of decision making over the substance of decisions made, especially when those decisions may violate universally recognized group member rights (Ramdas Reference Ramdas2006). Hundreds of millions of women suffer discrimination routinely through traditional or customary law systems. To cite an extreme case, more than a hundred million women—mostly in Africa—have suffered genital mutilation, according to a recent World Health Organization (2008) study. To cite another example, throughout Latin America, dramatic gender inequalities exist in land ownership and the holding of other resources (Deere and León Reference Deere and Magdalena2003). Despite such severe indicators of discrimination, political analysts point to improvements in women's participation in the holding of elected office worldwide as solid achievements (Inglehart and Norris Reference Inglehart and Norris2003; Reynolds Reference Reynolds1999), and they are. Furthermore, great merit may exist in multicultural approaches to governance, but caveats are in order.
This article argues that even in civic participation and governance, where discrimination would not seem as determinant as in sexual abuse and poverty, grave deficiencies in women's participation belie formal improvements in gender equality. Mexico, the nation studied here, offers a case in point as a new democracy where women are making steady, if insufficient, progress towards proportional representation in national political life (Baldez Reference Baldez2007; Rodriguez Reference Rodriguez2003). However, most work on political representation does not consider democracy's uneven spread, especially in Mexico's rural, poor, southern states.Footnote 1 Feminist defenders of indigenous autonomy in Mexico emphasize that gender discrimination in indigenous communities can be addressed within those communities, and that outsiders, such as Western feminists, human rights advocates, liberal individualists, or the Mexican state should not intercede (Forbis Reference Forbis, Eber and Kovic2003, 252; Hernández Castillo Reference Hernández Castillo2001, 24; Reference Hernández Castillo2005, 13–14; Marcos Reference Marcos and Marcos2005). The Zapatista insurgency in the 1990s, Mexico's most famous indigenous autonomy movement, claimed that men and women were caminando parejo, or “walking together” (Marcos Reference Marcos and Marcos2005, 87). That movement, by indigenous citizens in neighboring Chiapas state, has been credited—at least in part—with the Oaxaca State Legislature's passage in 1995 of the current law that allows indigenous communities to select mayors through customary law (usos y costumbres, or UC). This is any process they choose, but cannot involve political parties.
Although the debate between liberal pluralists and multiculturalists is polarized, a number of scholars argue for a third way (Benhabib Reference Benhabib2002; Ezeilo Reference Ezeilo and Marcos2005; Taylor Reference Taylor1994). These writers reject the assumption that equality before the law guaranteed by the national state has granted ethnic minorities and women equality. At the same time, these scholars are unwilling to accept a romantic or primordialist view of cultural groups as internally harmonious; that is, they recognize that women and other groups within groups may suffer internal discrimination. Accordingly, theorists ought to consider the rights of women and other disadvantaged groups within minority communities who may be granted special group rights. To this effect, in this article we focus on the case of Oaxaca, Mexico, where UC allowed citizens of designated municipalities to select their mayors via plebiscites, town assemblies, meetings of “town elders,” or other means, some of which clearly discriminated against women and other groups. This article seeks to argue for a limited multiculturalist “middle position” by considering the simple question of whether group rights regimes generally, and UC in Oaxaca specifically, are likely to more effectively protect women's rights than do the individual rights protections of national positive law (Ezeilo Reference Ezeilo and Marcos2005, 235).
Without seeking to dispose of the autonomy that UC recognition brings in many areas of public life, this article argues the normative position that individual rights standards must be adopted along with any communitarian rights of group recognition. While UC recognition itself may not be the cause of low female participation, statistical analysis of household survey data in UC and non-UC municipalities shows that women are far less likely to participate in local elections in UC municipalities. In the following section, we describe how the recognition of customary law was achieved in Oaxaca, and what this has meant in terms of women's rights. Then we present descriptive data showing that women participate less in UC communities, and we estimate three multivariate logit models demonstrating that this association generally holds when controlling for a number of intervening factors. We then consider case illustrations of the phenomenon, discuss its normative implications, and conclude.
OAXACA'S MULTICULTURAL EXPERIMENT AND ITS FORMAL AND DE FACTO EXCLUSION OF WOMEN
Shortly before the 1995 municipal elections in Oaxaca, Mexico, the state legislature voted to amend the state constitution, recognizing the right of three-fourths (73%) of the state's municipalities (36% of the state population; Velasquez Reference Velasquez2003) to autonomously govern themselves through usos y costumbres.Footnote 2 This occurred in the wake of the 1994 Zapatista uprising in neighboring Chiapas, and in a context of increasing political salience of indigenous identity throughout Latin America (Van Cott Reference Van Cott2005; Yashar Reference Yashar1999, Reference Yashar2005).
Under UC autonomy, local communities were permitted to exclude women, residents of outlying communities within the municipality (agencias), and citizens not born in the municipal “seat,” or population center, from full participation in the naming of authorities. According to official data, women were excluded from full participation in more than one in five (22.5%) UC municipalities (Eisenstadt Reference Eisenstadt2006), although some believe that many more municipalities denied women the vote but respondents refused to acknowledge this in anthropological surveys (Cruz Iriarte interview).Footnote 3 But even beyond the formal exclusion of women, which would not be legally possible in the absence of multiculturalist local autonomy regimes, statistical analysis of household-level survey data suggests that even where women were formally permitted participation, survey respondents in UC municipalities were much more likely to say that only men participated in local elections. While women's participation may have indeed been even more restricted before customary law was recognized, the first decade of customary law recognition had not notably improved conditions.
Many prominent multiculturalists acknowledge that group rights are impermissible if they discriminate against women or intragroup minorities (see for example, Hernández Díaz Reference Hernández Díaz and Díaz2007a, Reference Hernández Díaz2007b; Kymlicka Reference Kymlicka, Cohen, Howard and Nussbaum1999, 31). Accordingly, it might not be valid to dismiss their arguments based on the fact that under UC the state allowed local communities to disfranchise women, as they too might find this formal exclusion deplorable. But a more relevant question may be whether UC promoted or inhibited women's participation in communities where women were not already excluded under community norms. These might be the determinant cases in allowing us to really explore whether customary norms not directly related to gender discrimination nevertheless affect women's participation in local elections. Setting aside UC municipalities that exclude women, are local elections more likely to be viewed as male dominated in the formally inclusive UC municipalities when compared to non-UC municipalities? Multiculturalists might expect that under certain conditions, the community-based, consensus-building nature of customary elections promotes participation beyond the standard party system with secret ballots. UC critics might expect the opposite, that specifying restrictions on participation would limit the freedom of women to convey individual views, such as through secret ballots.
To address these empirical questions, we generated data from a survey of more than 1,800 individuals in 29 UC and non-UC municipalities throughout Mexico's southern state of Oaxaca, which is over 50% indigenousFootnote 4 and the home of over 15% of Mexico's estimated 12 million indigenous citizens (see Appendix for a full description of the survey methodology). Women possess full rights of participation in local leader selection in 13 of the 17 (76.5%) UC municipalities where individuals were surveyed (IEE 2003). According to 1997 data of all UC municipalities in Oaxaca, women possessed full voting rights in 74.2% of municipalities (see Table 1). The right of women to serve as mayors, however, is much less common. Women are eligible to serve in this post in only 3 of the 17 UC municipalities in the sample.
Table 1. Formal and perceived gender discrimination in Oaxaca municipal elections
*This is the dependent variable in the estimated statistical models.
Sources: The 1997 data are coded from Velásquez et al. 1997 by Eisenstadt and Viridiana Ríos Contreras. Information is for 411 of the 412 of Oaxaca's 570 municipalities originally designated as UC. Percentages do not add up to 100 because of missing information. Data on “Perceptions of Women's Voting” come from author tabulations of the 2004 survey used in this article's statistical analysis (valid N = 1,809 individuals in 12 non-UC and 17 UC municipalities). “Voting Rights” and “Right to Be Mayor” data come from the 2003 Oaxaca State Electoral Institute (IEE) information on UC practices for the sample UC municipalities only (N = 17 municipalities).
In addition, survey respondents were asked if men were the only participants in local leader selection. According to more than two-fifths (42%) of respondents in UC municipalities, women did not participate in local elections.Footnote 5 The corresponding response for those in non-UC municipalities is much lower, but certainly noteworthy at 15%. This simple finding reveals that beyond the official exclusion of women from political participation in more than one in four UC municipalities, the low-levels of women's political participation locally (whether due to informal exclusion or free preferences) appear to be quite prevalent, and are not absent in non-UC municipalities either. The low levels of women's participation in local politics is attributable to a range of factors—some of which we analyze in the statistical models explained in the following section. Beyond this analysis, our discussion of more than 50 open-ended interviews in Oaxaca (see the section “Applying the Theory”) reveals some of the ways in which women may be excluded or prefer not to participate despite the formal right to do so.
MODELING CAUSES OF GENDER DISCRIMINATION IN OAXACA
We now consider some possible explanations for this marked lack of female participation in local political decision making. We hypothesize two possible causes for the dependent variable of perceived discrimination against women by survey respondents, construct and evaluate three statistical models, and then assess evidence against author interviews. Why were informal beliefs affirming discrimination so much more pervasive than the actual institutional barriers against women? We offer a multicultural hypothesis and a liberal hypothesis.
Hypothesis 1: Multiculturalism, but with Patriarchy
The first hypothesis is that people will be more likely to perceive their localities as excluding women from participation in municipalities governed by UC. The argument here is that the formal institutionalization of UC leads to or facilitates the exclusion of women from political participation. In short, failure to reject this hypothesis is consistent with an affirmative response to Okin's question: “is multiculturalism bad for women?” Accordingly, we call this the multiculturalism, but with patriarchy hypothesis. Consistent with this hypothesis, Deborah J. Yashar (Reference Yashar1999) notes that a possible outcome of increased local autonomy, consistent with warnings of James Madison in the Federalist Papers, could be the emergence of illiberal fiefdoms.Footnote 6 Related to the present context, she notes the possibility that women's and other individuals' rights could be suppressed if they were seen to “threaten the sanctity of local autonomy and tradition” (Yashar Reference Yashar1999, 96). This hypothesis is also suggested from mainstream and “common sense” expectations, as reflected in comments by Patricia Espinosa, former director of the Mexican government's National Women's Institute, who, when expressing support for the 2001 Ley Indígena, noted the virtue that the legislation explicitly protected women's rights and would “rescue them from traditions and customs” (cited in Forbis Reference Forbis, Eber and Kovic2003, 237).
Changes in gender roles and progress toward political and social equality are often defined as characteristics of transitions from traditional to modern societies (Inglehart and Baker Reference Inglehart and Barker2000; Inglehart and Norris Reference Inglehart and Norris2003). Some argue that indigenous cultures are more patriarchal; accordingly, granting autonomy will tend to lead to negative outcomes for women (see discussion in Forbis Reference Forbis, Eber and Kovic2003, 236; see also Bartra Reference Bartra1997). Following from the perspective of liberal and modernization theorists, one would expect less participation among women in communities with higher percentages of indigenous language speakers or where other traditional practices, not equivalent to usos y costumbres elections or ethnicity, are prevalent. It is important to note that many UC municipalities do not have indigenous majorities (by language criteria) and many non-UC municipalities do (Velasquez Reference Velasquez2003, 25). Considering this important empirical fact together with the argument by these scholars that traditional cultures are more patriarchal, it should follow that women's exclusion is more prevalent in UC municipalities with higher percentages of indigenous residents. Insofar as most cultures are patriarchal, autonomy should positively correlate with exclusion, regardless of ethnicity. That is, self-determination is more likely to lead to unfettered discrimination in places that are relatively more patriarchal to begin with. If confirmed, the multiculturalism, but with patriarchy hypothesis would lend credence to liberal concerns that multiculturalist autonomy policies facilitate gender discrimination, even in cases where communities have not chosen to formally exclude women.
Hypothesis 2: Restoration of Traditional Harmony
The second hypothesis originates from the opposite end of the normative spectrum. This restoration of traditional harmony hypothesis will be strongly supported if recognition of customary law is associated with a lower likelihood of women's de facto exclusion from political participation. One explanation for this hypothesized pattern is provided by what Melissa M. Forbis characterizes as the romantic view of indigenous autonomy, which holds that gender discrimination in indigenous communities is a social pathology rooted in the history of colonialism, neocolonialism, social and political subjugation of indigenous communities, or some combination of these factors. Accordingly, the granting of local autonomy and the right to self-determination for indigenous peoples would allow for the restoration of gender harmony, and political and social harmony generally, to their communities (León Portilla Reference León Portilla1997, 9, as referenced in Forbis Reference Forbis, Eber and Kovic2003, 236).
Similar expectations are implied by indigenous rights advocate Sylvia Marcos, who recognizes the presence of gender discrimination in indigenous communities in Mexico, but argues that what outsiders often perceive as inequality between men and women may be in part due to a misunderstanding of indigenous cultural forms (2005). She notes, for example, the absence of the concept of equality, which implies stasis, in the Mesoamerican cosmovision. Indigenous women's struggle “toward a just relationship with their men” should be understood, rather, by reference to the metaphor of caminando parejo, or “walking together” (Marcos Reference Marcos and Marcos2005, 87). The process of walking together should occur within indigenous communities and between indigenous women and men. This constructivist perspective on gender and culture implies a theory about the process of gender relations and political change. According to this theory, then, the official recognition of usos y costumbres in Oaxaca might be expected, under circumstances of authentic autonomy, to free indigenous women and men to “walk together” with fewer external obstacles. The tacit assumption, at least with respect to the question of patriarchal social structures, is that the dominant culture and legal system of the Mexican state either perpetuates patriarchy or at least does not adequately protect women against discrimination, despite the presence of formally codified individual rights in the Constitution.
Whatever the veracity of claims that colonialism, racism, and social, economic, and political marginalization are responsible for exacerbating gender discrimination, it does not necessarily follow that recognition of autonomy would restore a presumed pre-Columbian harmony to indigenous communities or even, to return to Marcos's metaphor, open up space within which indigenous women and men can walk together.Footnote 7
As mentioned previously, many prominent scholars who argue from a multiculturalist or pro-autonomy perspective acknowledge that group rights are impermissible if they discriminate against women or intragroup minorities (see, for example, Hernández Díaz Reference Hernández Díaz and Díaz2007a, Reference Hernández Díaz2007b; Kymlicka Reference Kymlicka, Cohen, Howard and Nussbaum1999, 31). Taking the more simplistic spirit of the restoration of traditional harmony hypothesis, rather than the logical extension of its claims (which may lead to contradiction), we estimate separate logit models that exclude respondents who reside in communities where 1) women are formally excluded from voting in local elections and 2) women are ineligible to be elected as mayor.
According to the restoration of traditional harmony hypothesis, it is expected that autonomy can improve gender harmony in indigenous communities. However, insofar as the flexibility of UC institutions allows for greater manipulation by caciques — strong local or regional bosses—and state elites, these hopes of harmony may be dashed. This is particularly likely where, for whatever reason, women are more likely to support opposition to entrenched elites (see Anaya-Muñoz Reference Anaya Muñoz2004, Reference Anaya Muñoz2005; Eisenstadt Reference Eisenstadt2007; Eisenstadt and Ríos Contreras Reference Eisenstadt and Viridiana2008; Recondo Reference Recondo2007).
Although UC recognition itself may not be the cause of low female participation, statistical analysis of household survey data in UC and non-UC municipalities suggests that women are far less likely to participate in local elections in UC municipalities. This association holds when controlling for a number of intervening factors. Below we discuss the results of logit models of the dependent variable measuring perceptions of whether women participate in local elections. After defining the dependent and independent variables, we report statistical models and then analyze them in the next section (see Table 2 for descriptive statistics of the dependent and independent variables).
Table 2. Descriptive Statistics
Operationalization of the Dependent and Independent Variables
The following survey question is used for the dependent variable: “When local authorities are elected here, do only men participate?” We recode this variable to equal 1 when women are not perceived to be fully excluded—entailed by a “yes” response to the survey question—and 0 otherwise. The key explanatory independent variable is a dummy variable equal to 1 if the respondent resides in a UC municipality and 0 if he or she does not. A series of local- and municipal-level intervening variables are included as well. Finally, since the dependent variable measures individual perceptions of women's participation, we control for the respondent's gender and ethnicity, as well as whether he or she would be in favor of women governing locally (“discriminatory attitude”), in order to guard against the possibility that these factors might systematically bias responses about perceptions.
According to the hypotheses tested, perceptions about women's exclusion from local elections are expected to vary on the basis of whether or not the respondent resides in a UC community. To reject the restoration of traditional harmony hypothesis and support the multiculturalism, but with patriarchy hypothesis that presumes indigenous communities remain more discriminatory, we need to account for the ethnic composition of each community. To measure this, we include a variable for the percentage of the municipal population that speaks an indigenous language. We also tested models that include a variable to measure the “effective number of ethnicities” in the locality, but dropped it from the final models as it lacked statistical significance.
Control variables, such as one measuring the respondent's perception of the relative importance of the traditional local authority, “Local Authority Scale,” as contrasted with the state's government agents, and another attesting to the importance of “Traditional Practices” of community organizations are also included. We also control for the percentage of the surveyed population in each municipality that have discriminatory attitudes about women holding positions of local authority. These controls serve as important checks on the endogeneity problem, mainly that discriminatory practices may predate the formal adoption of UC laws in 1995. In other words, women might be more likely to face exclusion in UC communities not because of the recognition of autonomy but because municipalities identified by state legislators as UC municipalities did, in fact, already hold more traditional attitudes, and, accordingly, were already more discriminatory. Controlling for other measures of traditionalism helps to mitigate—though certainly not eliminate—this problem.
To measure “local authority,” we construct a composite variable that sums 9 individual dummy variables, adding one point to the index if the respondent replied that she or he would appeal to local authorities or the community assembly before the government on a range of issues.Footnote 8 Overall, this variable measures the level of importance of the community and local authority figures to address problems and conflicts, including the rape of a woman and women's health problems associated with childbirth, as well as agrarian conflicts and a number of other issues.Footnote 9 To measure the concept of “traditional governance,” we construct a 4-point scale (0 to 3) that allots 1 point if 1) a traditional community leader organizes the annual community festival, 2) if there is a requirement of unremunerated community work known as tequio, and 3) if there is a requirement that a person serve in positions of low-level public service, known as cargos, to qualify to be mayor.Footnote 10 The percentage of the municipal population holding discriminatory attitudes is constructed in two steps. A respondent's “discriminatory attitude” (=1 if discriminatory, =0 if not) is based on the following survey question: “Would you be in favor of a woman governing in this place?” Possible responses are “yes,” “no,” and “Don't know/Didn't answer.”Footnote 11 On the basis of this dummy variable, we calculate the percentage of the municipality (typically 66 survey respondents) showing discriminatory attitudes. Though imperfect, this variable can be seen as a proxy for the prevalence of patriarchal attitudes in the community.
We also include a variable that measures the percentage of local households headed by women, under the expectation that women will participate more in places where the traditional male-headed family is less prevalent. In these situations, women are much more likely to take on a public role as the family's representative in the community and are more likely to serve in lower-level cargos, which are often required in order to be eligible for higher levels of authority. Indeed, in various author interviews with women and men in two usos y costumbres municipalities in the Mixteca region with high rates of outward migration and high percentages of female-headed households (San Juán Mixtepec and San Miguel Tlacotepec), women had higher rates of public participation and were more likely to serve in cargos (in San Juán Mixtepec: interviews with Cruz Bautista 2007, Flores Reference Flores Cruz2007, Martinez Vasquez 2007; in San Miguel Tlacotepec: interview with Gutierrez Cortés 2007). Similarly, we control for the percentage of local population who, according to the 2000 Census, had migrated for work during the previous five years.
We also control for population, expecting that as population and urbanization increase, women will become increasingly engaged in political society (Inglehart and Norris Reference Inglehart and Norris2003) and patriarchal attitudes will decline. This is particularly important given the fact that UC municipalities tend to be rural and non-UC municipalities are more likely to be urban. We control for the level of poverty in the locality as a whole, as measured by the “marginalization index” of the Mexican census (CONAPO 2001). A variable measuring the percentage of the local population that were born in the given locality is also included (INEGI 2001; author calculations). The multiculturalism, but with patriarchy hypothesis might predict that places with higher percentages of individuals born in a locality would be more discriminatory, particularly insofar as this is an indicator of relative isolation from modern influences. To the contrary, the reconstituted traditional harmony hypothesis might predict the opposite, as original ties to the place of residence are a fundamental aspect of autonomy claims and can be expected to promote greater internal harmony by allowing women and men to walk together.
FINDINGS AND DISCUSSION
There is strong negative correlation (–0.29) between the dummy variable for women's exclusion (1 = not fully excluded) and UC residence (1 = UC). So respondents in UC municipalities are more likely to perceive full female exclusion from local elections. This relationship holds when estimating the first logit model in the analysis, Model 1. This model, (Table 3), contains all of the variables discussed previously and the universe of analysis is the entire sample of 1,831 individuals in 29 UC and non-UC municipalities. Models 2 and 3, also reported in Table 3, differ only in that they limit the case sets, excluding from the analysis residents of municipalities where women are formally excluded from voting (Model 2) and from serving as mayor (Model 3).
Table 3. Logit models of women's participation
*Significant at 0.1 level, **significant at 0.05 level, ***significant at 0.01 level, ****significant at 0.001 level.
Model 1 (Universe: all respondents) is highly consistent with the multiculturalism, but with patriarchy hypothesis, as despite numerous control variables, the perception exists that women are significantly more likely to be excluded from any participation in the selection of local authorities in UC municipalities (p < 0.001, see Table 3). Analyzing the predicted probabilities, holding all other variables at their means, residents of UC municipalities are 0.26 points less likely to perceive that women participate in local elections (see Table 4).
Table 4. Predicted Difference in Probability of Women's Participation High and low values of independent variables (significant at 0.01 level)
n.s. = Coefficient for variable was not statistically significant at the 0.01 level in the given model.
In Model 2, we exclude from the analysis all respondents from the three UC municipalities in the sample where women are officially excluded from participating. Although the effect of UC residence on likelihood of women's participation is somewhat less than in Model 1, the difference is small and the relationship is still negative and highly significant (p < 0.001). So this change in the case set does not markedly change this result, though the effect is slightly weakened (difference in predicted probability of 0.18, instead of 0.26).
When excluding from the analysis all but the three UC municipalities where women are eligible to be selected as mayor, the story changes significantly. In fact, when only this truncated sample is considered, the analysis shows that respondents living in these permissive UC municipalities, where women are eligible to serve as mayors, are significantly more likely to characterize their local elections as allowing women and men to compete, when compared to non-UC municipalities. Holding all other variables at their means, those in this small set of UC municipalities are 0.08 points more likely to answer that women compete when compared to their counterparts in non-UC municipalities.
Two reasons make it difficult to argue that these findings support the restoration of traditional harmony hypothesis. First, such a conclusion would be based on only a comparison among the 198 respondents from the three UC municipalities in the sample in which women have full political rights. Although the dependent variable varies within municipalities, this still may reflect the particularly open nature of the surveyed municipalities, rather than the enhanced harmony facilitated by UC autonomy. That said, the cross-sectional finding that women appear to participate at high rates in inclusive UC municipalities suggests the need for historical analyses to determine when and why women gained full political rights in these communities and not others. Second, the exclusion of such a vast majority (at least from the sample data) of municipalities from analysis begs an important question: If the case in favor of UC recognition and autonomy regimes rests on such a small number of “good apples,” has this position not essentially conceded the debate?
Not necessarily. The large number of UC municipalities in Oaxaca (418) means that even if a minority grant women full political equality, de jure and de facto, that minority would represent a large number of communities. Furthermore, if the statistical findings reported in Model 3 are reliable, we should be encouraged about the (conditional) potential of UC autonomy to enhance gender and multiculturalist harmony. And much to the contrary to the “perpetuated patriarchy” proponents' argument, Model 3 allows for the possibility that nondiscriminatory UC might even lead to more harmonious gender relations than in the more modernized communities and cities that elect their authorities on the basis of the system of political parties.
Before proceeding to a discussion of evidence from author interviews in UC communities in Oaxaca, we briefly discuss other noteworthy findings of this analysis. Variables that consistently had statistically significant and positive coefficients are discussed first. These include the percentage of the locality population that lives in a female-headed household (not significant in Model 3), the percentage of the population born in the locality (only in Model 1), and, somewhat surprisingly, the Local Authority Scale in the first two models (see Table 3 for coefficients and significance levels and Table 4 for predicted probabilities). One plausible explanation for the positive and significant relationship between the percentage of the population who live in female-headed households and attitudes toward women is that as women take on more authority at the household level, and as their real economic and social power increases, patriarchal attitudes about women's power and capacity for government diminish. Controlling for the other factors, among those living in localities with a very high percentage of female-headed households (58%), the predicted probability that community members perceive that women particpate was 43 percentage points higher than for those living in localities with very low percentages (3%, see Table 4) in Model 1 (difference of 0.28 in Model 2). The level of migration in the locality was not significant in any of the models, suggesting perhaps that this only has an effect on greater women's participation insofar as it results in a greater number of female-headed households. Finally, it is noteworthy that the percentage of the municipal population that is indigenous is not significant in the first two models. That is, challenging modernization theorists' claims, people in indigenous communities may not be more patriarchal. Furthermore, this variable is significant in the positive direction in Model 3, consistent with the restoration of traditional harmony expectations of autonomy proponents.
Variables that were negatively correlated with women's participation are now briefly discussed. Not surprisingly, the percentage of the sample in each municipality with discriminatory attitudes and the Traditional Practices scale variable were negatively related to perceptions of women's participation. This is important, as these variables help to mitigate against potential endogeneity problems. They are statistically and substantively significant across all three models (see significance levels and predicted probability estimates in Tables 2 and 3). The level of marginalization is not significant in the first two models, but is negatively correlated with perceptions of women's participation in the third model. Finally, the variable measuring whether the respondent speaks an indigenous language, included to control for any systematic difference in perceptions or interpretation of survey questions based on ethnicity or language, was significant (p < 0.001) and negative in Model 1 and Model 2, but not statistically significant in Model 3.Footnote 12
APPLYING THE THEORY: PATTERNS OF GENDER EXCLUSION IN UC COMMUNITIES
We now consider evidence from more than 50 open-ended interviews conducted with women and men in UC communities throughout Oaxaca since 2003. We picked the UC sites on the recommendation of academics, nongovernmental organization authorities, and public officials. We cannot claim to have been as systematic in selecting “illustrative” interviews as we were in constructing the representative survey sample (as explained in the Appendix). Interviews were not systematic, but rather were “soak and poke” ethnographic studies to validate our quantitative claims and fill in the patterns revealed by the survey data. This analysis, though anecdotal, suggests possible reasons for low levels of female participation in usos y costumbres municipalities and provides insights into the instrumental reasons that leaders may have for excluding women in certain cases. The municipalities focused on here were selected for study because they reflect a range of women's participation and formal and informal exclusion. In two of these municipalities, women are formally excluded but have individually or collectively pushed for rights of participation, and in one, women are formally included and actually participate at higher rates than men. Nevertheless, as will become evident, rights to participation for women can be seen as burdens, especially when this right is not accompanied with real decision-making power.
Challenges and Resignation in the Face of Formal Exclusion
Capulalpam de Méndez, a municipality of 1,313 residents in the Sierra Norte region, formally excludes women from participating in elections and from holding higher-level cargos (IEE 2003; Peréz Cosmés, interview, 2007). Though exclusion of women is official, it is important to note that UC institutions are typically not codified and change with a great deal of frequency (Alcántara interview, 2007). The flexible and contested nature of UC institutions is exemplified by the fact that former Municipal Treasurer Olga Toro is a very active participant in community assemblies and local political life generally. Nevertheless, she is an exception in Capulalpam as the only woman who currently participates. This low level of participation, she said, has to do with strong resistance of community males. Toro described one tactic that has historically been used by men to suppress women's participation in assemblies and meetings whenever they start to challenge prevailing roles. She described a period several years ago when women were starting to attend assemblies frequently and make inroads into local affairs. To suppress their active participation, anytime a woman attempted to speak, the men would make disruptive noise by vigorously shaking a cup or a can filled with pebbles (Toro 2007).
In 1998 in Asunción Tlacolulita (an indigenous-minority UC municipality of 699 residents that officially excludes women from voting), a group of women stormed the community assembly demanding to participate in the election process (Eisenstadt Reference Eisenstadt2006). Specifically, these women intended to vote for the opponent of the local cacique associated with the long-ruling Party of the Institutional Revolution (PRI). Follow-up field research reveals that despite this strong push for women's political participation, to this day women continue to be excluded from speaking or voting in community assemblies. A former leader of this group of women, Anastacia Zenón Flores, noted that the women are not pushing to participate in community assemblies now that things have calmed down and out of respect for custom and community unity (Zenón Flores interview, 2007). Nevertheless, Zenón Flores emphasized the learning process for the women involved in the protest movement in 1998, and she affirmed that if there were a repeat, the women of Asunción Tlacolulita would do the same thing, with greater confidence.
In an analytic case study of an usos y costumbres community in Oaxaca, Tad Muttersbauch (Reference Muttersbauch2002) finds support for increased women's participation in politics and public life among the opposition cooperativista faction and more resistance among the faction connected to the PRI, which ruled Mexico as a monopoly from 1929 until 2000. Furthermore, according to Zenobrio Cruz Sosa and Erminio Quiñones Osorio, activists in the leftist Party of the Democratic Revolution (PRD) from Asunción Tlacolulita, their group was supportive of women's participation during the 1998 electoral dispute, while PRI supporters opposed it (Cruz Sosa interview, 2007; Quiñones Osorio interview, 2007). Although concerns for equality were cited as perhaps the most important motivation for supporting women's participation, both men noted that the participation of women was a strategic benefit to their group because it increased their numerical, if not relative, majority.
In an interview with retired teacher and local PRI leader Viliulfo Ruíz Matias, this claim was not directly disputed. When asked about the participation of women in local leader selection, he noted that “I, as a professor, believe that women have a right to participate, but the people [read: the men] are not prepared to accept such a change in tradition” (Ruíz Matías interview, 2007).
As already noted, these cases elucidate the important fact that UC institutions are typically not codified into binding law and are often explicitly contested. This contested nature of UC institutions in Oaxaca has several important implications for the exclusion of women there. The fact that prevailing practices exclude women does not mean that women do not sometimes challenge these practices, as Toro continually and proudly does. It is easy to see how the flexibility of UC institutions opens up space for powerful groups to invent customary practices for instrumental reasons (Eisenstadt and Ríos Contreras 2008; Mamdani Reference Mamdani2002; Posner Reference Posner2005). One prominent example of this manipulation of custom is revealed by press coverage of 2001 postelectoral conflict in the Oaxaca municipality of Santa Catarina Minas. A new mayor was named by a majority vote in a General Community Assembly in October of 2001. The former mayor, whose preferred candidate had been defeated, protested to the Oaxaca State Electoral Institute (IEE, by its Spanish acronym) that the election was invalid, since women had participated in violation of traditional practices (Veléz Ascencio Reference Veléz Ascencio2001a,). In the end, the election of the mayor and cabinet was validated despite these challenges to the election's legitimacy (EDUCA 2002, 46; IEE 2003; Veléz Ascencio Reference Veléz Ascencio2001b). Despite the argument of the former mayor and his cadre, women have in fact participated actively in local elections since 1981, making this case an excellent illustration of the fact that custom is not infinitely manipulable by instrumentalist elites. The failure of Santa Catarina Minas elites to manipulate traditions, though, should not cloud the obvious point that UC flexibility is likely to invite political leaders to actively seek to change the rules to suit their objectives, whether progressive or regressive, populist or elitist.
Participating Without Deciding: Persistent Inequality Despite Formal Inclusion
When asked about women's participation in community assemblies to select local leaders, men and women typically responded that women did not want to participate. Customary requirements of assembly participation are often seen as burdens carried by adult men, who are usually required to participate in these activities, as well as in unremunerated administrative jobs (cargos) and manual labor (tequio). This work for the collective good takes people away from paid work in the labor market or from tending to their own household economies. Hence, women (and men) might prefer not participating in assemblies or in taking on cargos or tequios. As former Director of the Oaxaca State Electoral Institute Cipriano Flores Cruz commented, many women do not want to go to the assemblies to vote because they view it as a cargo (Flores Cruz interview, 2007).
Tomasa Martínez, a housewife and seamstress from the rural municipality of San Juán Mixtepec (population 7,423), noted that women increasingly make up majorities in the community assemblies that select municipal authorities and regularly serve in low-level cargos, though this was not the case historically. She attributes this increased public role of women to the high levels of outward migration of men from the municipality that began in the 1960s but has recently intensified. The greater likelihood of men's migration has often left women with the responsibility of doing their own traditional work in the household as well as the household and community duties traditionally fulfilled by their husbands. This includes serving in cargos, serving on public committees, and participating in community assemblies to select local authorities. That there have been changes in women's participation in public life in UC communities is also emphasized by human rights activist Sara Méndez (2007). Although women have gained power within communities through greater participation, deep contradictions remain and inequality persists. Echoing the sentiment expressed by Martinez, Méndez emphasizes the fact that this opening up is occurring in a context of profound rupture in the unifying logic of communities due to poverty, migration, and social exclusion that is by no means seen as entirely positive. This said, Méndez does not see a big difference in gender discrimination and limited women's political participation between UC and non-UC municipalities.
Taking into account this broader context of economic and family hardship, it is easy to see how this increased participation is a double-edged sword. On the one hand, participation in cargos and community assemblies consumes time and energy, at the expense of work in the household, tending to crops, and remunerated labor. On the other hand, even though women now have the right to participate and typically comprise a majority of participants in community assemblies, they still lack effective authority. As Martínez summed up, “we have the right to work, but not to decide.”
Given Martínez's circumstances, it is no wonder that, as Flores Cruz noted, women are often not interested in participating in assemblies or other communal jobs and authority positions. On the one hand, given the fact that women are unable to forgo their household responsibilities when they take on increased economic and political roles outside of the home, participation entails substantial net workload increase. As Martinez also noted, women's increased participation does not bring more power over the distribution of community resources or more influence over the decisions of the municipal authorities, making women's trepidations about participating understandable, even where they are not formally excluded. On the other hand, in situations like those of Martínez's San Juán Mixtepec, it is clear why men might tolerate increased roles by women, particularly when this does not diminish their own authority.
Discussion
This anecdotal evidence suggests that there are numerous possible economic and cultural factors that can limit women's participation and influence in local politics without a need for official exclusions.Footnote 13 The fundamental question implied is whether these low levels of participation have been exacerbated, improved, or unaffected by the official recognition of municipal autonomy in Oaxaca. Furthermore, can it be said that autonomous institutional structures such as those recognized in Oaxaca are generally good or bad for women?
Questions about how profound gender-related social and economic change is processed through UC institutional structures are of great interest. Is the presence of officially recognized customary institutions likely to insulate cultural, customary, and patriarchal social structures from change? Or is it possible that the recognition of UC provided communities with the autonomy necessary to empower themselves to collectively and more effectively deal with social and economic hardships? Although the statistical data in the previous section sheds some additional light on these questions, these changes are ongoing, and so the questions remain essentially open.
In summary, this section has suggested that 1) women's participation and political power can be limited due to numerous structural-economic factors; 2) participation in politics is often viewed by women as extra work with no tangible payoff in the form of increased influence; 3) entrenched elites such as PRI-connected caciques may have an instrumental interest in suppressing women's participation;Footnote 14 and 4) usos y costumbres institutions are fundamentally flexible, contested, and open to manipulation.
CONCLUSIONS
The preceding analysis reveals the complexity not only of Oaxacan politics but also of the debate between multiculturalist proponents of indigenous autonomy and liberal critics. Though this is ultimately a normative debate, it need not be purely deductive. Indeed, theorists in both camps, in addition to appealing to normative and philosophical principles of gender equality, may also lend credence to their point of view by appealing to the positive outcomes associated with policies consistent with their perspectives (and vice versa). Although the empirical findings of this analysis do not speak directly to the soundness of different normative philosophies of justice, they can serve as an important referent that theorists, philosophers, and policymakers might consider in further developing their arguments. Moreover, these findings serve as a strong caveat to policy analysts: While the tracing of advances in formal civic participation by women, such as through legislative or executive branch posts held, is important to document, much more basic discrimination still exists, in Oaxaca and wherever customary law discriminates. Basic impediments to the observance of human rights in these usually rural communities in the developing world are much less readily quantifiable, but they are in dire need of study.
Regarding the specific issues of women's political exclusion, customary law, and indigenous autonomy, the findings presented here suggest that uncritical advocates of the justice and purity of indigenous autonomy may have too romantic a perspective, as nominal UC autonomy may be exploited for instrumental reasons by local and state elites to the detriment of women's political rights. Insofar as UC institutions are more open to manipulation by local elites, or unfetter patriarchal tendencies of local leaders promoting a discriminatory status quo, autonomy might perpetuate patriarchy as well as authoritarian fiefdoms. This argument, strongly confirmed by Models 1 and 2, is consistent with claims by Todd Eisenstadt and Viridiana Ríos Contreras (2008) that Mexico's ruling political party at the time manipulated the establishment of UC municipalities (banning political parties in local elections) precisely to restrict and manipulate the electoral participation of opposition parties. Nevertheless, the analysis does provide ambiguous support for multiculturalists who suggest that when UC has disallowed the invention or perpetuation of formal discrimination against women, as in the three sample municipalities isolated in Model 3, autonomy does not seem to facilitate patriarchy. To the contrary, on the basis of this limited data, it appears that these municipalities may even, as predicted by the qualified “restoration of traditional harmony” hypothesis, open space for women and men to indeed “walk together.” Unfortunately for normative theory, the multiculturalists' optimism seems to hold only in this reduced sample of 3 out of 17 UC municipalities surveyed, where more tolerant attitudes regarding gender equality likely already existed. What can these findings tell us about the vast majority of communities in Oaxaca and in other customary law bastions where men and women walk together, but with the men in front, or where they are not even allowed to walk in tandem?
More research is needed to gain a better understanding of the variety of ways in which UC institutions function relative to non-UC institutions, particularly with regard to women's opportunities not only to participate in elections but also to equally share power in making important community decisions. As noted in the introduction, many prominent multiculturalists acknowledge that group rights are impermissible if they discriminate against women or intragroup minorities (Hernández Díaz Reference Hernández Díaz and Díaz2007a, 2007b; Kymlicka Reference Kymlicka, Cohen, Howard and Nussbaum1999, 31). The three sample municipalities in which women shared full political rights with men might be viewed as the models of autonomously recognized communities that these scholars really advocate. Based on the analysis in this article, we can concede that there is evidence that these limited multiculturalists are right, but only given conditions of full political equality such as in a scant 19% of our UC sample. The fact that this is a small minority of cases, however, presents the research question of how a few UC municipalities were able to achieve this level of equality and the policy question of how this can be achieved in the others. Clearly, recognition of formal autonomy is not sufficient for reaching the ideals of self-determination for historically marginalized groups and full political rights for those at risk of within-group marginalization.
This unprecedented survey and our analysis demonstrate that women are explicitly, and legally, excluded from full participation in voting and from holding positions of power in many UC municipalities (see Table 1), and they are excluded de facto in many more (recall that while women were allowed to vote in 26 out of 29 municipalities, a majority of respondents in 9 out of 29 sampled municipalities, 31%, answered that only men participated). In other words, while cataloging formal legal practices is important, informal practices of discrimination—in the form of norms of exclusion and/or active promotion of discriminatory acts by political bosses—are even more germane. And while norms of exclusion change slowly and, perhaps, only after a generation of education, formal institutions may change much more quickly. The Oaxaca experiment shows that the observation of human rights can be brought about by punctuated institutional change (such as the 1995 approval of customary law) where such ordinances reinforce norms of inclusion.
Mexico's formal national-level judicial institutions have barely begun to address these issues. Despite calls for a judicialization of rights, as per the courts' propulsion of the U.S. Civil Rights movement, Mexico's Supreme Court has taken no notable discrimination cases (Martínez interview 2008), and while the Electoral Court of the Judicial Power of the Federation has received a few appeals of Oaxaca voting rights violations, it has not yet set national policy through any ruling.Footnote 15 Although conflicts after usos y costumbres elections are often submitted to a special mediator in the Oaxaca State Electoral Institute, rather than to the state electoral court (where party-based election conflicts are resolved), no strict laws governing this mediation process exist, only the norms established in an incomplete “catalog” of usos y costumbres traditions described for each municipality.
Where norms of inclusion do not exist, such as in parts of Oaxaca, as well as in many of the world's other communities living under customary law, the trade-off between indigenous autonomy and liberal rights must be viewed more explicitly as such. Members of traditional communities do not always walk together when it comes to the rights of women and other minorities. Sometimes they walk apart autonomously, but sometimes they jostle and joust, at which time formal legal institutions protect the rights of women and other minorities, but only if the sojourners heed these institutions. And while such compliance is assumed in most studies, ours demonstrates that this is far from automatic, even in societies claiming widespread democratic participation.
APPENDIX: SURVEY SAMPLE AND VARIABLE CODING AND CONSTRUCTION
The survey was designed by Todd Eisenstadt, Araceli Burguete of the CIESAS-Sureste anthropology Institute, and María Cristina Velásquez, affiliated with the CIESAS-Isthmo, and administered in Spanish and six indigenous languages, between June 2002 and Feburary 2003. Eisenstadt devised the sampling technique, which sought to identify a representative sample of respondents over 18 years of age in at least three urban census tracts (“AGEBs” in Mexico) and/or rural communities (“localidades”) per municipality in 22 to 30 municipalities each in Chiapas and Oaxaca, and Zacatecas (the present analysis only focused on the Oaxaca portion of this survey). Burguete, Velásquez, and Francisco Muro of the State Autonomous University of Zacatecas trained interviewers, most of whom were bilingual (Spanish and indigenous languages) university students. The survey overrepresented indigenous respondents by sampling 50% from indigenous-majority communities and 50% from non-indigenous-majority communities in Chiapas and Oaxaca, but randomly generated the other strata (statistical tracts, blocks, and households for municipalities with more than 4,000 residents, and localities and resident lists for localities with fewer than 4,000 residents, for which statistical tracts are not kept) based on 2000 population data. Several substitutions of municipalities and submunicipal localities were required. The overall margin of error is less than ±2%, although it can be higher within particular question categories. This margin of error holds with a 95% confidence interval. For more on the survey, see Eisenstadt Reference Eisenstadt2009.
Note that in one of the municipalities in which we code women as having formal participation rights, San Juán Tepeuxila, only married women are allowed to participate. We did not code this as discriminatory against women because only married men are allowed to participate as well. Indeed, there are many other groups that face limited rights in some UC municipalities, including unmarried men and women, those who were not born in the municipality, those who do not live in the municipal seat, and non-Catholics. Also, as we state in the text, this coding of discrimination was based on the reporting of local officials (rather than direct observation by coders). These local authorities may have exaggerated their claims to make their communities seem less objectionable to outsiders, but no other source presently exists to make this assessment.
Dependent Variable
“When local authorities are elected here, do only men participate?” (1 = “no,” 0 = “yes”)
Independent Variables
1. Usos y Costumbres.
2. Percent Indigenous Municipality.
3. Traditional Practices (0 = no indicator or traditional governance, 3 = all measured indicators). The component questions (equal to 1 or 0) were added together to create the composite scale. The component survey questions were the following:
a. Who organizes the annual community festival? (1 = mayordomo, the traditional community patron of the feast, 0 = church, a committee of festivals or the government)
b. Is community work, known as tequio, required of community members? (1 = yes, 0 = no)
c. Does a person have to serve in a cargo, a low-level public service post, in order to become mayor? (1 = yes, 0 = no)
4. Local Authority (0 = no indicators of community/assembly authority primacy, 9 = all measured indicators of traditional government). The following component questions were added together to create the Community Authority Scale variable. One point was added to the scale for a given respondent if the response was “Authority/Assembly/Community” as opposed to “Government”:
a. “What authority does one appeal to first to resolve the theft of a cow or other livestock?”
b. “What authority does one appeal to first to resolve the ‘looting’ of wood?”
c. “What authority does one appeal to first to resolve a land conflict?”
d. “What authority does one appeal to first to resolve the environmental contamination of the forest?”
e. “What authority does one appeal to first to resolve womens' health problems after giving birth?”
f. “What authority does one appeal to first for the education of the children?”
g. “What authority does one appeal to first for help getting a business loan?”
h. “What authority does one appeal to first to address the rape of a woman?”
i. “What authority does one appeal to first to resolve questions of access to a well/water source?”
5. Percent Discriminatory. This variable is constructed by taking the percentage of the population in each surveyed municipality that answered negatively to the following question:
“Would you be in favor of a woman governing in this place?”
6. Percentage Female-Headed Households, Locality.
7. Percent Migrants. Percent of locality residents who have migrated and returned during the past five years.
8. Natural Logarithm of Local Population.
9. Marginalization Index (min = –2.06, max = 1.50). This indicator is estimated by the Consejo Nacional de Poblaciόn (CONAPO) of Mexico based the XII Censo General de Población y Vivienda, 2000, INEGI. The nine indicators are 1) percentage of population 15 years or older who are illiterate; 2) percentage of population 15 years or older who have not completed primary education; 3) percentage of private home occupants without drainage or exclusive sewer service; 4) percentage of private home occupants without electricity; 5) percentage of private home occupants without running water (entubada); 6) percentage of private homes with some level of overcrowding; 7) percentage of private home occupants with dirt floors; 8) percentage of population in localities with fewer than 5,000 residents; and 9) percentage of employed population with income of less than two minimum wages. Methodology document can be accessed on the Internet at: http://www.conapo.gob.mx/publicaciones/indices/pdfs/006.pdf.
10. Percent Born Locally.
11. Gender (1 = female, 0 = male).
12. Discriminatory. Indivudal's response to question: “Would you be in favor of a woman governing in this place?” (1 = “no,” 0 = “yes”).
13. Indigenous. If the respondent speaks an indigenous language = 1, if not 0.
