英语论文网

was found that a total of 1615 out of 2471 respondents were currently married; of them 789 resides in slum areas. The analysis is based on this sub-sample.

The function estimated will take the following form:

Contraceptive choice = F ( Socio-religious identity, culture, age of respondent and its square, Education of respondent and her partner, log of wealth index score, place of last delivery, number of living sons, number of living daughters, participation in economic activities)

Most studies (D'Souza 2003, Kamaal 2000, 2007, Kamaal and Huda 2008, Stephenson 2004, Waiz 2000) use binary or multinomial logistic models to identify the determinants of contraceptive prevalence rates. Now, the choice variable analyzed is discrete but not binary - possible responses are not using any method, using folkloric method, using traditional method or using modern method. Given the low incidence of folkloric methods, traditional and folkloric methods can be clubbed together, so that the choice variable can assume three values (no method, traditional or folkloric, modern). In such cases a multinomial model is appropriate. Further, given that the options can be ordered ordinally in terms of reliability, the ordered logit model is appropriate (McCullagh 1980).

The ordered logit model is based on the proportional odds, or parallel lines, assumption. As this assumption is strong, Brant (1990) has suggested the use of a method to test the validity of this structure. If tests indicate that the assumption of proportional odds is violated, the question arises whether the coefficients vary for all explanatory variables, or only some (Williams 2006). In this context, the Likelihood Ratio test is used to select the parsimonious model.

调查结果-3. Findings

Analysis of DHS-3 data indicates that contraceptive prevalence rates in Calcutta's slums (72 percent) are higher than those in the other 8 cities covered in the survey (64 percent) and also national levels (56 percent). Only 11 percent had never used contraceptives.

Analysis of the methods used indicates that the incidence of female sterilization is very high, followed by use of condoms and pills (Fig. 2). This corresponds to the preference for terminal methods observed in other Indian slums (Das and Shah 2001, Agarwal and Bharti 2006). Sen (2001) and Chattopadhyay et al. (2004), too, make a similar observation for Calcutta's slums. The reasons underlying preference for a method that allow less freedom than condoms pills, IUD, etc. in terms of birth spacing needs investigation. But what is interesting is that 60 percent of respondents who have sterilized had a boy child on their last delivery. This may be indicative of the desire to satisfy son preference while limiting family size.

Analysis of the reasons for not adopting any contraceptive shows that factors like religious strictures, opposition by partner or family members are unimportant. About 63 percent of the respondents were either infecund, sub-fecund or had undergone hysterectomy. This implies that contraceptives are viewed primarily as a method for family planning, rather than protection against STD or AIDS/HIV. A further 11 percent either do not have intercourse or do so infrequently. Thus about 73 percent of the non-users do not require family planning methods. Examination of future intentions of non-users reveals that about