ÕÅÔÚÐÂ

Source: Project Muse Scholarly Journals Online

Social Forces 82.3 (2004): 1063-90.

Symbolic Gender Politics, Religious Group Identity, and the Decline in Female Genital Cutting in Minya, Egypt*

Kathryn M. Yount

Emory University

Abstract

The introduction of female genital cutting to Egypt predates the arrival of Christianity and Islam. Elsewhere, a belief that the practice is religiously significant has justified its continuation, and a belief that it contradicts religious tenets has instigated its abandonment. Findings from Minya, Egypt, show more rapid declines in the prevalence of female genital cutting and more negative effects of maternal education on the odds of circumcising daughters among Christian compared to Muslim families. Such differences have emerged as Islamists have engaged the state in public debates over women¡¯s authentic roles and as Christian voluntary organizations have adopted alternative ¡°gender symbols¡± as indicators of group identity in public discourses on development.

Although the origins of female genital cutting1 are uncertain, scholars believe that its introduction to northeastern Africa predates the arrival of Christianity and Islam (Assaad 1980; Gruenbaum 2001; Mackie 1996, 2000). According to Hosken (1993), signs of excision were detected among female Egyptian mummies of the sixteenth century B.C., and Herodotus found ancient Egyptians practicing excision on his visit in the fifth century B.C. One theory holds that the center of the distribution of female genital cutting was the Meroite civilization of the Middle Nile, which was prominent from the second millennium B.C.E. to the beginning of the first millennium C.E. (Adams 1977; Mackie 2000). Scholars argue that, within the region, trade of female slaves who were genitally cut initiated discourse about the fidelity-promoting purposes of the practice, which spread within one group as a convention sign of marriageability before spreading to overlapping groups (Mackie 1996, 2000).

Today, female genital cutting is practiced among various religious groups globally (Population Reference Bureau [PRB] 2001). In five northern provinces in Sudan, 98% of Muslim women compared to 1.2% of (mainly Coptic) Christian women were circumcised in 1982 (El Dareer 1982). At the same time, female genital cutting is not practiced in 80% of the Islamic world, including Saudi Arabia, Jordan, Iran, and Iraq (Gordon 1991; Joseph 1996) and occurs among Coptic Christians in Egypt, several Christian groups in Kenya, and the Falasha Jews of Ethiopia (Gordon 1991; Gruenbaum 2001; Hosken 1993; Joseph 1996; Lane & Rubenstein 1996; Obermeyer 1999). Among the Sara, one of Chad¡¯s largest ethnic groups, the highest rates of female genital cutting occur among rural Catholics (96%) and the lowest rates among urban Protestants (53%) (Leonard 1996). Among the Sara of Myabe village in Chad, adolescent girls reportedly initiated the practice only two decades ago, citing ¡°neither religion nor ¡®tradition¡¯ to bolster their claims that ¡®female circumcision is good¡¯¡° (Leonard 2000:181). Female genital cutting is more prevalent among non-Muslims than Muslims in Tanzania and Kenya and among Muslims than other religious groups in Burkina Faso, the Central African Republic, Egypt, Eritrea, Mali, and Sudan (PRB 2001).

The pre-Judeo-Christian roots of female genital cutting and its widespread prevalence among various religious groups globally refute a popular belief that the origins of the practice lie in formal religious doctrine (e.g., Gordon 1991). Contemporary interpretations of religious tenets are cited as reasons to adopt or continue the practice, however, and beliefs that the practice reflects religious or ethnic ideals provide a means to assert group identity. Among Mandinga men and women in Guinea Bissau, the three ethnic groups that practice female genital cutting are Muslim, and the practice is considered to be a cleansing rite that enables Muslim women to pray in the proper fashion (Johnson 2000). In Sudan, religion has been one of the most commonly mentioned reasons among men and women to favor the practice (El Dareer 1982), and two groups living near each other in Sudan have considered different forms of female genital cutting to be important markers of ethnic identity: the Zabarma women previously of West Africa believe that clitoridectomy is religiously correct, whereas their Arab-Sudanese neighbors believe that infibulation is religiously superior (Gruenbaum 1991).

Just as local interpretations of religious doctrine are used to rationalize continuation of the practice, alternative interpretations are used to justify minimizing or abandoning it. Some Islamic practitioners in Sudan have advocated less severe forms of the practice on religious grounds (Gruenbaum 1991, 2001), and religious prohibition has been among the most commonly mentioned reasonsthat northern Sudanese men and women reject the practice (El Dareer 1982). ¡°Respondents said it was contrary to . . . Islam because it is mentioned neither in the Koran nor the Hadith, and is not practiced in many other Muslim communities¡± (El Dareer 1982:79). Among the 46 women who opposed the practice in Leonard¡¯s (1996) study in Chad, 40% cited a religious injunction against it, and all but one of these women were members of the local Protestant church. ¡°Through the denigration of all ¡®traditional¡¯ practices, including . . . female circumcision, the [Protestant] church is attempting to bring about a radical shift in perspective¡± (Leonard 1996:262). The ability of the Protestant church to affect change has been limited to its membership, however, since local Catholic clergy are more tolerant of the practice.

The above discussion suggests that formal religious affiliation may be less relevant than local religious ideology as a determinant of change in the practice of female genital cutting. To develophypotheses about patterns of change in this practice among religious groups in Minya, Egypt, I draw on the theoretical work of Anthias and Yuval-Davis (1989) regarding the participation and representation of women in ethnic group processes. Anthias and Yuval-Davis (1989) argue that ¡°gender symbols¡± like female genital cutting and others that concern the sexual behavior of women serve to reproduce religious boundaries and to signify religious differences when used strategically in public discourse. These authors argue further that this process of differentiation may arise in a context in which representatives of one group lobby the state to reinforce certain customs and religious norms through the passage of national legislation. Applied to the case of Egypt, I show that as Islamists attempted during the 1970s and onward to assert national political power by engaging the Egyptian state in debates over women¡¯s ¡°authentic¡± roles in the private sphere, local Christian voluntary organizations began to promote ideals and activities that reflected a different vision of women¡¯s empowerment and included the eradication of female genital practices.2 I anticipate that religious tensions arising from the unequal institutional relationships of Muslims and Christians to the state likely hindered the diffusion of these competing ideas about women in public discourses on development, instigating greater declines in the prevalence of female genital cutting among Christians than Muslims during this period. I anticipate also that support among Christian voluntary organizations for women¡¯s empowerment in the public and private spheres catalyzed negative effects of higher maternal education on the odds of circumcising daughters. In the following section, I draw on historical context to substantiate the above hypotheses regarding the interplay between symbolic gender politics, religious group identity, and female genital practices in Egypt.

Symbolic Gender Politics, Religious Group Identity, and Female Genital Practices in Egypt

Egypt is home to the largest percentage of Sunni Muslims and the largest number of Orthodox Christians in the Middle East. Constituting some 6% to 18% of the total population (Ibrahim 1996; Zeidan 1999),3 Coptic Christians are spatially concentrated in the governorates of Assyut, Sohag, and Minya in Upper (southern) Egypt, where some 60% of all Copts live and where they comprise around 20% of the population in that region (Zeidan 1999). Historically, Christians and Muslims alike have practiced female genital cutting; the latest estimates of national prevalence suggest that the practice remains nearly universal among ever-married Egyptian women of reproductive age (El-Zanaty & Way 2001).

Relations between Christians and Muslims, each consisting of diverse subgroups (Hatem 1994; Ibrahim 1996; Kurzman 1998), have varied historically, and have been reconfigured and renegotiated asboth groups have adapted to changing political and economic circumstances (Ibrahim 1996; Zeidan 1999). As a consequence, scholars have variously characterized the relationship of Christians to the larger body of Muslims in Egypt as ¡°equal,¡± ¡°protected¡± (dhimmi ), and ¡°persecuted¡± (Armanios 2002; Nikolov 2002; Zeidan 1999).4 It is not the task of this article to explicate historical shifts in Muslim-Christian relations in Egypt, but rather to focus on selected events of the nineteenth and twentieth centuries that are significant to explain potential variation by religious affiliation in the practice of female genital cutting.

According to Shukrallah (1994), ¡°modernization¡± became associated early on in Egypt with colonialism and Western influence. The Napoleonic invasion and Muhamed Ali¡¯s rise to power in 1804, for example, aligned Egypt ideologically with Western ideals of ¡°progress,¡± which practically involved expansion of the military and industry, ¡°modernization¡± of the educational system (including the founding of a school of midwifery in 1831-32 and the granting of scholarships for Egyptians to study in Europe), and the state¡¯s assumption of control over Al Azhar, an Islamic institution (Badran 1991; Shukrallah 1994). Economic decline and the subsequent colonization of Egypt by Great Britain initiated advocacy for reforms among the Egyptian ruling classes, intellectuals, and religious leaders, and a social movement for ¡°modernization¡± merged with a nationalist movement against British occupation. Shukrallah (1994) argues further that secular and Islamic trends developed from this movement, both of which initially saw ¡°progress¡± as consistent with Western ideals of modernity and women¡¯s liberation.

Under Gamal Abdel-Nasser¡¯s regime, religion and secularism were tied to its political ideology and agenda (Shukrallah 1994). On the one hand, efforts to expand and secularize the public sectors of education, health, and employment brought new groups intothe new nation¡¯s identity and citizenship. On the other hand, Arab de facto secularism has distinguished itself from Western de jure secularism insofar as the latter has called for formal separation of church and state whereas the former has to some degree recognized Islam as the religion of society either to solidify political alliances or to face Islamist political challenges (Hatem 1999; Hermassi 1993). Shukrallah (1994) argues that the decision of Nasser¡¯s regime not to secularize all sectors of civil society led to a ¡°denial of full citizenship¡± (21) to women and Christians, and she illustrates this point by contrasting codes of personal status that continued to allow coresident men to forbid women to work with new laws that gave women equal opportunities for education and paid employment. According to Shukrallah (1994), ¡°[T]his denial of full citizenship for both categories of citizens can be seen as an early form of what would be used . . . in the later Muslim revivalist movement¡± (21).

The growth of Islamic revivalism and the Egyptian state¡¯s strengthening of its Muslim identity emerged more clearly following Anwar Sadat¡¯s project of economic liberalization (infitah ), which led to rapid increases in prices, unemployment, inflation, and a ¡°crisis of faith¡± among some groups in socialist and capitalist pathways to ¡°modernization.¡± The state under Sadat promulgated a constitution that made Islam the official state religion, politicized the role of Shari¡¯ah law, and articulated a dichotomy between women as public citizens and private family members governed by Shari¡¯ah law (Hatem 1994, 1999; Shukrallah 1994). For example, although the constitution of 1971 stated that ¡°citizens are equal before the law . . . in public rights and duties, with no discrimination made on the basis of race, sex, language, ideology, or belief,¡± it simultaneously guaranteed a ¡°balance . . . between a woman¡¯s duties towards her family . . . [and her] equality with man in the political, social, and cultural spheres . . .without violating the laws of the Islamic Shari¡¯ah¡± (Badran 1991:222). The balance that was struck in the early 1980s became clear when Islamic revivalists successfully pressured the High Constitutional Court to overturn reforms to the Personal Status Law that had been instituted in 1979. Whereas reforms curtailed privileges previously held by men in the private spheres of marriage, divorce, and child custody, the Personal Status Law of 1985 compromised certain of these changes. One compromise involved removal of the assumption of injury occasioned by a polygamous marriage and addition of the requirement for a wife to establish having suffered ¡°harm¡± from her husband¡¯s polygamous union in order to obtain a divorce. A second compromise pertained to the requirement that a divorced wife in custody of minor children have exclusive rights to the marital home for as long as she retained custody (unless her former husband provided another dwelling). Although the requirement to provide accommodation for the custodial mother remained in the 1985 legislation, the former husband gained exclusive rights over this dwelling.5

These debates over women¡¯s rights in the private sphere exemplify a strategy among Islamists to regulate family life as a means of exercising national political power and of asserting a distinct Muslim national identity (e.g., Hegland 1999). Revivalist Islamic discourses in particular have emphasized ¡°cultural authenticity¡± in opposition to Western cultural influence and the ¡°moral and ethical structure of society versus the structural insecurities caused by modernization¡± (Baykan 1990 in Shukrallah 1994:17). Like the discourses of non-Muslim and Muslim groups elsewhere (Chhachhi 1989; Jalal 1991), a major site of this symbolic politics is gender, whereby women come to represent cultural authenticity and bear responsibility for transferring traditional identity (Anthias & Yuval-Davis 1989; Kandiyoti 1991; Shukrallah 1994). One of the best-known gender symbols is the practice of veiling, which itself is not a political act but becomes one when its meaning is publicly debated. ¡°Control of the symbol¡± becomes tied to control of the state (Sedra 1999:220).

A similar ¡°frenzy of interpretation¡± arose in Egypt over female genital cutting after the 1994 International Conference on Population and Development (ICPD).6 After a broadcast by CNN depicting a lay practitioner¡¯s circumcision of a young girl, the mufti of Egypt publicly declared that female genital cutting has no foundation in the Koran. Following this declaration, Sheikh Gad al-Haq Ali of Al Azhar issued a fatwa (religious ruling) that ¡°female circumcision is a part of the legal body of Islam and is a laudable practice that does honor to the women¡± (Kassamali 1998).7 Others sought to strengthen the foundations of the practice by calling upon custom as a source of Islamic law (Abu-Sahlieh 1994).8 Opposition to support for eradication motivated the minister of health to seek advice on the subject from an expert medical committee, which issued statements about the dangers of procedures that are performed by untrained practitioners (Seif El Dawla 1999). On these grounds, the minister of health decreed in late 1994 that doctors could circumcise girls in designated facilities and at fixed times and prices, claiming that medicalization of the practice would reduce complications and eventually end the practice. Subsequent pressure from international agencies, as well as the reported deaths of girls who were circumcised in hospitals, instigated a renewed ban on the practice in public hospitals (which was overturned and then reinstated in late 1997). Such contestations highlight the way in which customary and revivalist Muslims have engaged the Egyptian state in ¡°symbolic gender politics¡± as a means to assert a national Muslim identity in opposition to transnational influences.

Internal responses to the increasingly Muslim character of the Egyptian state during the last 25 years have been at least twofold. Most widely publicized has been periodic but increasing levels of sectarian strife, particularly in Upper Egypt, where Coptic Christians are more numerous (Ibrahim 1996; Nisan 1991). Less widely publicized but no less significant has been the proliferation of religious-based social services and private voluntary organizations, including private religious schools and health facilities that are affiliated with mosques or churches (Ibrahim et al. 1996; Nikolov 2002; Shukrallah 1994; Sullivan 1994). Whereas Shukrallah (1994) argues that the increasingly religious character of previously secular public institutions has created a ¡°crisis of identity¡± (24) for women and Christians, Nikolov (2002) argues that local Christian services are ¡°part of the larger mission of the Church to bring the stray . . . back into the ecclesia¡± (3), and Zeidan (1999) asserts that the growth of Coptic voluntary organizations (especially in Upper Egypt) is a ¡°marker of Coptic activism¡± (60) (see Sullivan 1994).9

Several Christian voluntary organizations have endorsed the ideals of comprehensive community development, women¡¯s public and private empowerment, and the rights of children, and several such organizations have initiated activities to eradicate female genital practices. Sullivan (1994) notes that the Christian community also is known for its ability to seek assistance from international sources. The Coptic Evangelical Organization for Social Services (CEOSS), for example, began as a literacy project in the 1950s. In 1960, CEOSS was registered with the Egyptian government and charged with serving Muslim and Christian communities. CEOSS separated from its parent organization ¡ª the Egyptian Evangelical Church ¡ª and undertook activities in Beni Suef, Minya, and metropolitan Cairo. Its programs grew to include agriculture, income generation, education, infrastructural development, and health. In 1976, CEOSS established the Family Life Education Unit, which identified female genital cutting, early marriage, and ¡°bridal deflowering¡± as customs that were harmful for women.10 CEOSS has helped establish women¡¯s committees in the villages in which it has worked as part of a ¡°deliberate effort¡± to empower women (Tadros 2000:26). As of 1994, CEOSS received 75% of its funding from Europe, the U.S., and Canada and at one point was the only private voluntary organization to be registered with the U.S. Agency for International Development (Sullivan 1994).

Similarities between the mission and implementing strategies of CEOSS and those of other Christian voluntary organizations in Egypt are striking. The Bishopric of Public, Ecumenical, and Social Services of the Coptic Orthodox Church (BLESS), for example, was established in 1962 to provide comprehensive services to the poor in Cairo and included among its goalsthe eradication of female genital practices (Nikolov 2002). As early as 1974, the bishopric developed a comprehensive program to expand the services that it offered to women and to develop the skills of women and girls, and its strategy for implementation explicitly identified women¡¯s participation as integral to the process of development. As of 2000, BLESS¡¯s campaign against female genital practices has been undertaken in 24 communities throughout Egypt (Nikolov 2002). The missions and activities of these and other Christian organizations arguably reflect a wider trend within the Christian social service community to adopt an integrated vision of women¡¯s public and private empowerment as a marker of religious identity (see Sullivan 1994 for other examples).11

The previous discussion suggests that efforts of customary and revivalist Muslims to engage the Egyptian state in symbolic gender politics as a means to assert an ¡°authentic¡± national Muslim identity created an obstacle to the efforts of liberal Muslims to alter popular support for female genital practices. Although Christians were marginalized from public debates over the meaning of this and related gender symbols in Shari¡¯ah law, the proliferation of social services of Christian origin since the 1960s provided a public space in which Christian identities could be enacted and by which Coptic Christians in particular could address social problems ¡°according to the values and norms of the Coptic Orthodox tradition¡± (Nikolov 2000:2). Not facing the same ideological conflict over female genital cutting within their leadership (Abdel-Hadi n.d.), Christian activists were able to form state-approved voluntary institutions that operated under a politically acceptable umbrella of nonsectarian community development while promoting a vision of women¡¯s empowerment that differed from revivalist Islamic discourses on gender. This alternative vision emphasized the integrated empowerment of women in the public and private spheres and included the eradication of female genital practices.

I therefore anticipate that the adoption by Christian voluntary organizations of these ideals and related activities fostered declines in the prevalence of female genital cutting among Christians before its practice in Egypt emerged on the international political agenda in 1994. Adoption of a unified view of women¡¯s empowerment in the public and private spheres by some such organizations also likely catalyzed negative effects of higher maternal education on the odds of circumcising daughters. By contrast, I expect that reactionary views of some Muslims regarding women¡¯s public and private roles may have stalled declines in the prevalence of female genital cutting and similar effects of higher maternal education on its practice within the wider Muslim community. To test these hypotheses, I use population-based data representative of seven districts in Minya governorate to compare trends in the practice of female genital cutting among daughters of Christian and Muslim families, beliefs about the effects of circumcision and noncircumcision among Christian and Muslim mothers of these daughters, and the influence of maternal education on perceived effects of the practice and the odds that Christian and Muslim daughters are circumcised.

The Setting of the Minya Governorate

Minya governorate is a relatively poor area in Upper Egypt located about 200 kilometers south of Cairo and extending about 80 kilometers along the Nile. The cultivated (and thus habitable) land is about 30 kilometers wide at the widest point and lies mostly on the west bank of the river. The capital city is Minyatown (population about 1 million), and the governorate houses several large district towns and rural villages. Some of the villages on the east bank of the Nile are quite isolated from health facilities. About 20% of the residents of Minya are Coptic Christians, and most remaining residents are Muslim. The economy is largelyagrarian, levels of fertility and mortality are high compared to levels in governorates in Lower Egypt, and formal educational attainment continues to be particularly low among women. Despite evidence of a decline in the overall prevalence of female genital cutting in Minya, the probability that girls aged 10-14 years are circumcised is still .75 (Yount 2002). Secondary or higher education among ever-married mothers aged 15-54 years in Minya is associated with markedly lower odds of circumcising a daughter, and adjusted odds of circumcising a daughter are 4.5 times higher among Muslims than Christians in the governorate (Yount 2002).

Since the adoption of Islam as the state religion in Egypt, Minya has been the site of periodic and increasing levels of sectarian conflict (Ibrahim 1996; Nisan 1991). Before and during this period, Minya also has been the site of efforts by religious and secular voluntary organizations to promote comprehensive community development and to curb the practice of female genitalcutting (Abdel-Hadi n.d.; Abdel-Tawab & Hegazi 2000). In 1982, CEOSS founded a committee of 12 female representatives of all churches in the predominantly Christian village of Deir El Barsha (Abdel-Hadi n.d.). Members of the committee received training to improve their skills, to supervise projects, and to engage in consciousness-raising to prevent practices including female genital cutting (Abdel-Hadi n.d.). In 1995, CEOSS initiated an intensified educational program in 22 communities in Minya governorate that focused on girls at risk of circumcision (aged 7-13 years) and their mothers (PRB 2001). CEOSS established committees in these villages that included a village leader (omda ), a sheikh, and a priest, and coeducational implementing teams from CEOSS lived with designated families, registered all girls aged 7-13, and introduced them to topics related to health, literacy, and eventually female genital cutting. These activities reportedly reduced the rate of female circumcision in 8 of the 22 communities (World Health Organization [WHO] 1999), but more so in homogeneous, Christian villages (PRB 2001). Recently, some Catholic organizations have undertaken activities to reduce the prevalence of female genital practices in Minya (Masterson & Swanson 2000), and other organizations, including the Upper Egypt Society and CARITAS Egypt, have implemented programs to raise awareness of the effects of female genital practices (Abdel-Hadi n.d.).

Despite indications that religious boundaries may increasingly demarcate levels of female genital cutting in Minya, no research has compared systematically and with attention to the historical political context attitudes toward female genital cutting, levels of various types of cutting, and trends and determinants of change in the prevalence of the practice among Muslims and Christians throughout the governorate. The present analysis is intended to fill this gap.

Data

Questionnaires for this study were implemented in conjunction with the Two Governorate Linkages Survey (TGLS), a five-round longitudinal study of child morbidity and adult women¡¯s reproductive behavior that was undertaken in Qaliubia and Minya governorates during 1995-97 (Langsten & Hill 1996). In each governorate, eligible participants from a representative sample of approximately 3,000 households were interviewed at three-month intervals, and information about household composition and characteristics of ever-married women aged 15-54 years was collected. A women¡¯s status module was implemented in Minya during the fifth round of the TGLS, from which part of the data for this study is drawn.12

In structured interviews with all ever-married women aged 15-54 in identified households, local interviewers collected information about their sociodemographic characteristics (e.g., marital status, pregnancy and contraceptive histories, education) and characteristics of their husbands and households. Survey respondents alsoanswered closed-ended questions about their own circumcision status (whether or not circumcised, type of circumcision, age at circumcision, and attendants) and open-ended questions about the perceived effects of circumcision and noncircumcision. For each resident or nonresident daughter who was at least five years of age at the time of the respondent¡¯s interview, respondents completed a circumcision history that included questions about the current age and circumcision status of daughters, mother¡¯s intention to circumcise uncircumcised daughters, and age at circumcision and person who performed the procedure for already circumcised daughters.13

Dependent Variables

Dependent variables in this analysis include whether a daughter at least five years of age was circumcised, whether she was excised,14 and whether the mother of an uncircumcised daughter intended at the time of the interview to have her daughter circumcised. Also analyzed are perceived effects of circumcision (alternatives identified, bleeding, scarring, desired by religion, good for girl, satisfies husband, normal or tradition, cleansing or purifying, beautifying, other positive effect, some positive effect, no effect)15 and perceived effects of noncircumcision (marital problems, problems with fertility, health problems, psychological problems, unattractive or unfeminine, unclean, not good for girl or girl¡¯s reputation, excessive sexual behavior or commission of a ¡°sin,¡± other problems, should not imitate foreigners, required or a norm).

Independent Variables

Independent variables of interest in this analysis include religious affiliation of the family (Christian = 1, Muslim = 0)16 and maternal educational attainment (none [reference], primary or preparatory, secondary or higher).Control variables in the analysis are those that have been associated with attitudes and behavior pertaining to female genital practices in Egypt (El-Gibaly et al. 1999, 2002; Yount 2002), including age group of the daughter in years (5-9 [reference], 10-14, 15 or older), birth order of the daughter among living sisters (1 [reference], 2, 3, 4 or higher), maternal circumcision status (not circumcised [reference], circumcised), paternal educational attainment (none [reference], primary or preparatory, secondary or higher), residence (rural [reference], urban), and household assets (none [reference], 1, 2 or more).17

Analytic Methods

In descriptive analyses, I compare the sociodemographic characteristics of Muslim and Christian residents in Minya. I also compare relative frequencies of the circumcision status of daughters, type of attendant at circumcision among circumcised daughters, and maternal intent to circumcise uncircumcised daughters, by religious affiliation of respondents. Standard errors of these estimates are adjusted to account for nonindependence of the circumcision experiences of biological sisters in the sample (N = 3,212 daughters, 1,776 circumcised, 1,424 uncircumcised).

I then use life table techniques to compute age-specific probabilities of circumcision for age cohorts of daughters, by religious affiliation. Let ¦Ói be individual failure or censoring times and aggregate these data into intervals tj, j = 1, . . ., J and tJ+1 = ¡Þ. Let dj and mj be the number of failures and censored observations that occur during the interval tj and Nj the unadjusted number alive at the start of the interval tj. Using actuarial methods to adjust for censored observations, the adjusted number at risk at the start of the interval is nj = Nj - mj/2. The product limit estimate of the survivor function is

and the cumulative ¡°failure¡± time is Gj = 1 - Sj (StataCorp 2001).18 Finally, I compare relative frequencies of responses to open-ended questions about the perceived effects of circumcision and noncircumcision among Christian and Muslim mothers of daughters in the sample, adjusting standard errors of these estimates to account for the stratified, cluster-sample design (Rao & Scott 1981, 1984).

In multivariate analyses, I use logistic regression to estimate unadjusted and adjusted odds ratios and 95% confidence intervals (Christian vs. Muslim) that daughters are circumcised or excised (N = 3,212), that mothers intend to circumcise uncircumcised daughters (N = 1,424), and that mothers perceive certain effects of circumcision and noncircumcision (N = 1,504). A robust variance estimator is used to estimate standard errors of coefficients in models based on the sample of daughters to account for nonindependence of the circumcision outcomes of biological sisters, and robust standard errors are estimated for coefficients in unadjusted and adjusted regression models based on the sample of daughters¡¯ mothers to account for the stratified, cluster-sample design. All adjusted models include maternal age, maternal education, paternal education, household assets, and urban-rural residence, and adjusted models for daughter¡¯s circumcision status also include daughter¡¯s age, birth order among living sisters, and maternal circumcision status. Finally, interaction terms between religious affiliation and maternal education are added to selected full main effects models to assess the extent to which effects of maternal education on attitudes andbehavior vary by religious affiliation of the family.

Results

In addition to the higher rates of secondary or higher education among Christian than Muslim mothers in the younger age cohorts, a higher percentage of daughters of Christian than Muslim families were not circumcised at the time of interview (54% vs. 42%), and a lower percentage of daughters of Christian than Muslim families had experienced more extensive forms of cutting (30% vs. 43% excised) (Table 2). A lower percentage of Christian than Muslim mothers of uncircumcised daughters also intended at the time of interview to have their daughters circumcised (54% vs. 85%), such that the percentage of daughters already or expected to be circumcised was lower among Christian than Muslim families (75% vs. 94%, assuming that mothers who ¡°did not know¡± their intentions would not circumcise their daughters, and 81% vs. 96% assuming that mothers who ¡°did not know¡± their intentions would circumcise their daughters). Except for small differences in the percentage of circumcised daughters who had their procedure performed by a health barber or tamargi (male nurse), there are nodifferences by religious affiliation in choice of circumciser for daughters who were circumcised at the time of interview.

Figure 1 compares the cumulative probabilities of circumcision among age cohorts of Christian (left) and Muslim (right) daughters. Estimation of 95% confidence intervals for point estimates permits visual assessment of significant differences in these cumulative probabilities by age group. As shown, the cumulative probability of circumcision is lower for each successive age cohort of daughters, regardless of religious affiliation. Declines in this probability are greater among daughters of Christian than Muslim families, however. For example, the cumulative probability of circumcision by age 10 is .68, .65, .47, and .28 among Christian daughters aged 25 or older, 20-24, 15-19, and 10-14, respectively. Although the cumulative probability of circumcision by age 10 among Muslim daughters aged 25 or older is similar to that among Christian daughters aged 25 or olderat the time of interview (.73), and although this probability is lower for each successive age cohort of Muslim daughters, the decline is less pronounced among Muslim than Christian daughters (.70, .64, and .49 for daughters age 20-24, 15-19, and 10-14, respectively). Among cohorts aged 25 or older, 20-24, and 15-19, the cumulative probability of circumcision by age 13 (e.g., the likely end of the ¡°risk period¡± for Egyptian girls) declines from .98 to .88 and .68 among Christians but remains relatively constant among Muslims (.95, .95, and .93, respectively). Therefore, although the probability of circumcision appears to be declining among both religious groups in Minya, a much larger percentage of Christian than Muslim daughters can expect to remain uncircumcised by age 13. A majority of Christian and Muslim daughters continue to be circumcised in Minya, however.

Table 3 summarizes the responses of daughters¡¯ mothers to open-ended questions about the perceived effects of circumcision and noncircumcision, by religious affiliation. In general, Muslim mothers perceive more often than do Christian mothers that female genital cutting has positive effects: 17% versus 7% respond that the practice is good for the girl, 34% versus 21% suggest that it is cleansing or purifying, and 81% versus 68% state that the practice has some positive effect. A higher percentage of Muslim than Christian mothers indicate that the practice is associated with bleeding, and a lower percentage of Muslim than Christian mothers state that the practice has no effect ¡ª findings that are consistent with the higher percentage of Muslim than Christian daughters experiencing more severe forms of cutting.

Regarding the perceived effects of noncircumcision, some mothers believe that such a decision would lead to marital problems, problems with fertility, excessive sexual behavior or commission of a ¡°sin,¡± health problems, psychological problems, or other problems and that it would be unattractive or unfeminine, unclean, or not good for the girl or the girl¡¯s reputation. Some mothers state that the practice is required or is a norm and that one should not refrain from circumcising a daughter simply to imitate foreigners. A marginally higher percentage of Muslim than Christian mothers indicate that not circumcising a daughter would be unattractive or unfeminine (9% vs. 5%), and a higher percentage of Muslim than Christian mothers report that it would not be good for the girl or the girl¡¯s reputation (8% vs. 3%) or would lead to undesirable or excessive sexual behavior (22% vs. 14%). Only Muslim mothers indicate that not circumcising a daughter is disadvantageous because one should not imitate foreigners ¡ª although few Muslim mothers offered this response.

Table 4 shows unadjusted and adjusted odds of a daughter¡¯s circumcision status and maternal perceptions of the effects of circumcision and noncircumcision among Christian and Muslim families. Results are consistent with expectations. The unadjusted odds that a daughter is circumcised or experiences a more severe form of cutting (excision) and that a mother intends to circumcise uncircumcised daughters are consistently lower among Christian than Muslim families. After adjusting for other variables, the magnitudes of these odds ratios are reduced even further: the adjusted odds that a daughter is circumcised are .27 (95% CI .19-.39) times lower and that a daughter is excised are .41 (95% CI .30-.57) times lower among Christian than Muslim families. The adjusted odds that mothers intend to circumcise uncircumcised daughters are .18 (95% CI .11-.28) times lower among Christian than Muslim families.

Although not all are significant, the unadjusted and adjusted relative odds that a mother reports specific perceived effects of circumcision are in directions that are consistent with the relative odds of circumcision, excision, and intended circumcision. Namely, the adjusted odds of reporting that bleeding is an effect of circumcision are .56 times lower among Christian than Muslim mothers, and the adjusted odds of reporting that circumcision has no effect are 1.73 times higher among Christian than Muslim mothers. Christian mothers also have significantly lower adjusted odds than Muslim mothers of reporting that circumcision is good for the girl (OR = .38, 95% CI .22-.66), is cleansing or purifying (OR = .50, 95% CI .32-.78), or has some positive effect (OR = .50, 95% CI .27-.91).

Regarding the perceived effects of not circumcising a daughter, Christian mothers have marginally lower unadjusted odds than Muslim mothers of reporting that noncircumcision is unattractive or unfeminine; however, the adjusted odds that Christian and Muslim mothers report this effect are not significantly different. Christian mothers have .34 (95% CI .16-.71) times lower adjusted odds than Muslim mothers of reporting that noncircumcision is bad for the girl or her reputation and .59 (95% CI .40-.85) times lower adjusted odds than Muslim mothers of reporting that noncircumcision leads to excessive sexual desire or the commission of a ¡°sin.¡± In sum, Christian daughters have lower adjusted odds than Muslim daughters of being circumcised, of experiencing more severe forms of cutting, and of having their mothers report an intention to circumcise, and Christian mothers have lower adjusted odds than Muslim mothers of reporting selected positive effects of circumcision and selected negative effects of noncircumcision.

Table 5 shows predicted probabilities (and their 95% confidence intervals) of circumcision among daughters and predicted probabilities that their mothers report some positive effect of circumcision, by religious affiliation. These probabilities are based on adjusted logistic regression models (Table 4) that also include an interaction between maternal education and religious affiliation. In the case of a daughter¡¯s circumcision, predicted probabilities are computed for first daughters aged 15 years or older with mothers aged 35-44 years who have no formal education and are circumcised, with fathers who have no formal education, and who live in rural households that own no major assets. Estimated coefficients of this model (not shown) indicate a marginally significant interaction between religious affiliation and secondary or higher education of the mother: the adjusted odds that daughters of mothers with secondary or higher education versus daughters of those with no education are circumcised is .23 times lower among Christians than Muslims. This significant interaction is apparent in the sharp decline in predicted probabilities of circumcision among daughters of Christian mothers with secondary or higher education compared to daughters of Christian mothers with less education (.47 vs. .92 and .88, respectively). By contrast, there is no difference in these predicted probabilities among daughters of Muslim mothers with secondary or higher education and those of Muslim mothers with less education (.92 vs. .97 and .97, respectively).

In the case of maternal reports that female circumcision has some positive effect, predicted probabilities are computed for mothers aged 35-44 years who have no formal education, with husbands who have no formal education, and who live in rural households that do not own any major assets. Estimated coefficients of this model (not shown) indicate a highly significant interaction between religious affiliation and secondary or higher education of the mother: the adjusted odds that mothers with secondary or higher education compared to mothers with no education report that female genital cutting has some positive effect is .30 times lower among Christians than Muslims. This significant interaction is apparent in the sharp decline in predicted probabilities of reporting some positive effect of circumcision among Christian mothers who have secondary or higher education and Christian mothers with less education (.48 vs. .80 and .76, respectively). By contrast, there is no difference in these predicted probabilities by educational level among Muslimmothers (.81 vs. .87 and .84, respectively).

Conclusions

This article adds to feminist theory on the strategic use of gender symbols to articulate religious group identity by describing how transnational political discourse and evolving institutional alliances of majority groups with the state can condition the gender ideals and actions of minority groups (Anthias & Yuval-Davis 1989; Sedra 1999; Shukrallah 1994). This article also adds to feminist research on this topic by undertaking the first population-based comparison of maternal attitudes about and levels and determinants of female genital cutting among Muslims and Christians in Egypt. Findings are consistent with expectations and show that daughters of Christian families in Minya have lower adjusted odds of circumcision, lower adjusted odds of experiencing more severe forms of cutting, and lower adjusted odds of having their mothers report an intention to circumcise them if they were uncircumcised at the time of interview. Declines in the age-specific probability of circumcision also have been substantially more dramatic among Christians than Muslims and suggest that abandonment of the practice has been under way among Christians in Minya since at least the late 1980s. Christian mothers also have lower adjusted odds than Muslim mothers of perceiving customarily positive effects of circumcision and customarily negative effects of noncircumcision. Negative effects of higher maternal education on the odds that daughters are circumcised and that mothers report some positive effect of circumcision are substantial among Christians and absent among Muslims.

These findings are consistent with the argument made at the outset of this article that popularization during the last 25 years of an Islamic ideology identifying women as the keepers of ¡°cultural authenticity¡± and ¡°traditional identity¡± may have hindered declines in the prevalence of female genital cutting and potentially negative effects of women¡¯s higher education on the odds of circumcising daughters among Muslim families in Minya. By contrast, sustained opposition to ¡°traditional practices¡± among Christian voluntary organizations may have encouraged rapid and dramatic declines in the prevalence of female genital practices among Christians. Finally, the strong negative effect of higher maternal education on female genital cutting among Christians is consistent with the argument that the promotion by Christian voluntary organizations of an integrated view of women¡¯s empowerment in public and private life may have catalyzed negative effects of women¡¯s formal education on the odds of circumcising Christian daughters. Growing sectarian tensions in Minya during this period due in part to unequal relationships of Christians and Muslims to the state also may have limited the diffusion of competing ideals about women and development and contributed to the growing divergence by religious affiliation in the prevalence of female genital cutting in the governorate.

Notably, other environmental factors may have had direct, indirect, or synergistic effects on emerging differences in the prevalence of female genital cutting in Minya (Abdel-Tawab & Hegazi 2000), and views about the practice may not have been homogeneous within religious groups (Meinardu 1967 in Lane & Rubenstein 1996). Nevertheless, marked declines in the prevalence of the practice among Christians suggest that activities and ideologies favoring decline spread among Christians as Islamists were advocating a revival of women¡¯s ¡°traditional identity.¡± Therefore, the findings of this study are consistent with the idea that variation in the prevalence of customary practices like female genital cutting in Egypt may be understood in the context of historically unequal relations of religious groups to the state and their strategic use of competing gender symbols in national and transnational discourses on development.

Future research should include population-based studies of the patterns and determinants of female genital cutting in other Egyptian governorates where faith-based organizations are working to eliminate the practice. Also needed are experimental design studies that examine the impact in different religious communities of comprehensive, community development programs that incorporate activities to improve women¡¯s position in the public and private spheres. Comparisons of change in the practice of female genital cutting in Egypt and other customary practices elsewhere would permit assessment of the cross-cultural generalizability of this hypothesized interplay between symbolic gender politics, religious or ethnic group identity, and practices that reflect the position of women in society. Finally, since the passage in 1997 of national laws prohibiting the practice of female genital cutting in Egypt, national levels of support for it have reportedly declined (El-Zanaty & Way 2001). Whether this new ideological backdrop encourages abandonment of female genital cutting among all Muslims and Christians remains a worthy question for future research.

*Comments from Frank Lechner on the manuscript and advice from Sunita Kishor, Kenneth Hill, and Ray Langsten during the development of the circumcision history and the fieldwork are greatly appreciated. The data for this analysis were collected while the author was a doctoral student in the Department of Population and Family Health Sciences at the Johns Hopkins School of Hygiene and Public Health. The fieldwork for this study would not have been possible without the support of Hoda Rashad and staff at the Social Research Center of the American University in Cairo. Assistance from Rebecca Warne Peters in the formatting of this manuscript is greatly appreciated. A grant from the Sociology Program of the National Science Foundation (SBR-9632340) supported the study design and data collection, and the Departments of International Health and Sociology at Emory University supported the finalpreparation of the manuscript. Direct correspondence to Kathryn M. Yount, Ph.D., Department of International Health and Sociology, Emory University, 1518 Clifton Rd., NE, Room 724, Atlanta, GA 30322. E-mail: kyount@sph.emory.edu.

Endnotes

1. The World Health Organization (1997) uses the term ¡°female genital mutilation¡± to refer to all types of female genital practices. Here, I use the terms ¡°female genital cutting¡± or ¡°female genital practices¡± for this purpose. Yount and Balk (forthcoming) provide a detailed rationale for this choice of terminology.

2. Several variants of Islam exist in Egypt. I use the terms ¡°revivalist,¡± ¡°customary,¡± and ¡°liberal¡± Islam in accordance with the definitions proposed by Abu-Sahlieh (1994) and Kurzman (1998). Customary Islam is characterized by a combination of regional and shared practices. Each region of theIslamic world has forged its own version, and some Muslims have tried to strengthen the foundation of practices like female genital cutting by calling upon custom as a source of Muslim law (Abu-Sahlieh 1994). Revivalist Islam, also known as Islamism or fundamentalism, opposes local deviations of Islam, emphasizes the illegitimacy of local political institutions, privileges revivalists as the unique interpreters of Islam, and calls for an ¡°authentic¡± Muslim identity that opposes Western influence. The several variants of Liberal Islam commonly critique customary and revivalist traditions for being ¡°backward¡± and argue that Islam is compatible with Western liberalism (Kurzman 1998). Liberal interpreters of Islam argue that discrimination on the basis of gender in Shari¡¯ah law is unacceptable and that Islamic law must be reconciled with universal human rights (An-Na¡¯im 1998).

3. The true percentage of the Egyptian population that is Coptic Christian is uncertain. Government figures from the 1976 census suggest that the correct percentage is 6.3, whereas Coptic sources indicate that the true figure is closer to 20. Zeidan (1999) and Ibrahim (1996) offer detailed discussions of this debate.

4. Use of the term ¡°minority¡± with reference to the Christian population in Egypt is highly contested. Where I use the term, I do so to refer to differences in the relative sizes of Muslim and Christian populations in Egypt.

5. Amendment to the Law of Personal Status in January 2000 gave women more options for divorce.

6. The International Conference on Population and Development was held in Cairo, Egypt, on September 5-13, 1994. Organized jointly by agencies of the United Nations (UN), the conference convened over 11,000 registered representatives of governments, UN specialized agencies and organizations, intergovernmentalorganizations, nongovernmental organizations, and the media. Over 180 states participated in drafting a Programme of Action, which focuses on the association between population and development and on meeting the needs of individuals rather than on achieving demographic targets. The acknowledged key to this new approach is the empowerment of women and the provision of choices through expanded access to education, skill development, employment, health services, and policy- and decision-making processes. Among the actions identified to address the needs of girls is the elimination of ¡°female genital mutilation.¡± Paragraph 4.22 states: ¡°Governments are urged to prohibit female genital mutilation . . . and to give vigorous support to efforts among non-governmental and community organizations and religious institutions to eliminate such practices¡± (United Nations Fund for Population Activities [UNFPA] 1994).

7. A mufti is a religious leader in Islam who deals with questions on Islamic jurisprudence. In Egypt, the president appoints the mufti, who fills the top governmental post for a religious scholar. The Sheikh of Al Azhar heads the Supreme Council of Al Azhar and represents the highest religious authority within Sunni Islam. A fatwa is a legal statement in Islam that may be issued by a mufti or a religious lawyer in cases where the legal rulings of Muslim scholars are undecided or uncertain.

8. Historical banning of the practice in Egypt reportedly arose due to uncertainty about whether the Hadith (a collection of the sayings and acts of Prophet Muhammad) considers the practice to be sunnah (a duty) or makrama (an embellishment) (Abu-Sahlieh 1994; El Dareer 1982; Gordon 1991; Gruenbaum 2001; Lane & Rubenstein 1996).

9. According to Ibrahim et al. (1996), the number of Egyptian private voluntary organizations (PVOs) grew from 65 in 1900 to 12,832 in 1990. Others estimate that there were approximately 14,000 total registered private voluntary organizations by 1990 (Sullivan 1994). Sullivan (1994) argues that although Islamic PVOs have extensive services, Christian PVOs are more active per capita and more efficient in the generation of revenue and expenditures.

10. ¡°Bridal deflowering¡± refers to a range of practices that are performed to demonstrate the sexual purity of a new bride, including the breaking of the hymen by a midwife on the wedding night and the public display of a bloodstained cloth. Such tests have been important in some settings to preserve the honor of the new bride and her natal family (Antoun 1968).

11. Two caveats to this argument are noteworthy. First, Armanios (2002) argues that the construction of gender identity has for centuries been central to the Coptic church¡¯s construction of a Coptic identity but that the role of women in this project was in part to uphold certain patriarchal ideals. Armanios (2002) writes that the ¡°threat of eradication as a ¡®race¡¯ was never far from the minds of Copts. . . . Thus a significant emphasis was placed not only on the reproductive responsibilities of all Copts (mainly directed at women), but also on the literal breeding of a specific type of woman. A woman was not only to bear children but must herself be intelligent, proficient, and skilled to bring up the future sons of the nation¡± (114). Thus, enhancement of the skills and educational attainment of Coptic women has in part served to maintain men¡¯s authority in the Coptic family and community. A second caveat is that views on female genital cutting have varied historically across Christian denominations in Egypt; however, leaders in the Coptic Christian community have either been silent on the practice or have favored its abandonment (Abdel-Hadi n.d.).

12. Round 5 occurred from October 1996 to January 1997, when public debate about the legal status of female genital cutting was under way. Notably, these debates may have biased responses to questions about intent to circumcise daughters who were uncircumcised at the time of interview.

13. The attitudes of Christian and Muslim fathers with regard to female genital cutting are unknown because fathers were not interviewed. However, even though the origins of the practice are believed to lie in patriarchal views about women¡¯s sexuality and marriageability, studies show that mothers are often the immediate decision makers about the type and timing of a daughter¡¯s circumcision (e.g., Yount 2002). In addition, this analysis controls for paternal education, and results (not shown) suggest that paternal education is not associated with the odds of circumcising a daughter and that the effects of paternal education do not vary by religious affiliation.

14. Excision usually involves removal of the clitoris and labia minora rather than clitoridectomy or removal or the labia minora only. See WHO (1997) for descriptions of standard types of female genital cutting.

15. The Arabic word for ¡°bleeding¡± that was used in the questionnaire (naziif ) is a general term, the connotations of which vary according to the context. With reference to female genital cutting, the term may have positive or negative connotations depending upon whether the practice itself is perceived to be positive or negative. With reference to a wedding night, the term may have positive connotations in places where proof of a bride¡¯s purity is valued, but the term may have negative connotations when used in reference to an accident that causes someone to bleed. ¡°Some positive effect¡± includes the following responses: desired by religion, normal or tradition, beautifying, good for girl, satisfies husband, and cleansing or purifying.

16. During data collection, questions that would distinguish people of different Christian denominations were not asked. However, Shari¡¯ah rules of personal and private law state that a Muslim man may marry a Christian or Jewish woman but that a Christian or Jewish man may not marry a Muslim woman. According to An-Na¡¯im (1998), this rule derives from sura 4, verse 141, of the Qur¡¯an, which states that a non-Muslim may never exercise authority over a Muslim. In addition, a Muslim may neither inherit from nor leave inheritance to a non-Muslim. For these reasons, intermarriage of Christians and Muslims is believed to be rare in Minya, Egypt.

17. Assets about which questions were asked include a residential building other than the dwelling unit, commercial or industrial buildings, private car, motorcycle, bicycle, transport equipment (truck, van, bus, etc.), farmland, other land, livestock(horses, goats, sheep, etc.), poultry, farm equipment (tractors, etc.).

18. This formula is a Kaplan-Meier product-limit estimator with an actuarial adjustment and aggregation of events into interval lengths of one year. The actuarial adjustment accounts for censored observations in the product-limit formula by use of the equality nj = Nj - mj/2, and deaths are assumed to occur evenly over each interval, as indicated by the straight lines connecting estimated proportions circumcised at each year of age.