In Kenya, women's decisions to have their daughters circumcised are associated with various individual and community-level characteristics, such as education, media exposure and ethnicity, according to an analysis of data from a national survey.1 Overall, 38% of women reported experiencing genital cutting, and 46% of circumcised women already had or planned to have their daughters circumcised. In multilevel regression analyses, the likelihood that a mother would decide to have her daughter circumcised decreased as her level of education and exposure to media rose, and increased with the proportion of circumcised women in the community. In addition, the likelihood that a daughter would be circumcised varied significantly by the dominant ethnic group in the community.

To assess the extent to which individual and community-level characteristics are associated with women's decisions to have their daughters circumcised, the analyst used data from the 1998 Kenya Demographic and Health Survey (DHS). For the descriptive analyses, the total DHS sample of 7,873 women aged 15-49 from 530 clusters was used; for the multilevel logistic regression analysis, the sample was restricted to the 1,406 circumcised women from 354 clusters whose oldest daughter was younger than age 20.

Overall, 38% of women aged 15-49 reported experiencing genital cutting. The proportion of women who had been circumcised increased with age and decreased with education and exposure to media. The relationship between genital cutting and household wealth was less clear: Greater proportions of women in the third and fourth lowest quintiles of wealth (50% and 47%, respectively) than of those in the bottom quintile (31%) or top two quintiles (26-30%) had been circumcised. Forty-two percent of rural women reported having experienced genital cutting, compared with 23% of urban women. Finally, the proportion of circumcised women varied dramatically by ethnic group: For example, 97% of Kisii and 89% of Masai were circumcised, compared with 43% of Kikuyu and only 1% of Luo.

Among circumcised women, 46% planned to circumcise their daughters or had done so. Patterns in daughters' circumcision by education, media exposure and household wealth were similar to those for women overall; however, the proportion circumcised did not vary as much by area of residence or by ethnicity among daughters as it did among women overall. Also, no clear pattern was seen in the relationship between daughter's circumcision and daughter's age.

In the first logistic regression model, which included only individual characteristics, mothers' level of education and media exposure were negatively associated with the likelihood of deciding to have their daughters circumcised; daughters' age and household wealth were nonsignificant. When community characteristics were added to the model, urban residence was positively associated with the likelihood of daughters'circumcision. In addition, household wealth in the cluster was associated with an increased likelihood and mean media exposure in the cluster was associated with a decreased likelihood of daughters' circumcision. Mothers' level of education remained significant, but the level of media exposure became nonsignificant.

In the final model, which added variables for community norms regarding circumcision, being part of a cluster in which the dominant ethnicity was Kalenjin, Kamba, Kikuyu, Meru/Embu or Mijikendaa/Swahili was negatively associated with the likelihood of daughters' circumcision, whereas being part of a cluster where the dominant ethnicity was Kisii was positively associated with the practice. In addition, for every one-point increase in the proportion of circumcised women in the cluster, the likelihood of daughters'circumcision increased by nearly 3%. Level of mothers' education and urban residence remained significant, level of media exposure became significant again, and both the mean education of women and the mean household wealth in the cluster became nonsignificant.

According to the researcher, the data suggest that "women's behavior resembles that of other women [who] live near them" and "respondents' decisions to circumcise their daughters vary significantly by ethnic group." She concludes that although individual influences on circumcision are possible, "decline in the prevalence of female genital cutting is likely to take place simultaneously across social groups, rather than as a result of isolated individual decisions."—J. Rosenberg


1. Hayford SR, Conformity and change: community effects on female genital cutting in Kenya, Journal of Health and Social Behavior, 2005, 46(2):121-140.