An Application of Ordinal Regression Models in Analysis of Women Fertility Count Data in Kalburgi District of Karnataka State, India: A Case Study

Abstract

The aim of the study is to use the application of ordinal regression models in determination of determinants of urban and rural women fertility in Kalburgi, Karnataka, India. A total of 1500 women were selected in which, 750 were from rural and 750 were from urban area by using systematic random sampling procedure. The data on different determinants were collected through well prepared questionnaire with direct personal interview method. The data were analyzed with ordinal regression model with logit, probit, clog-log and nlog-log built link functions using statistical software SPSS 21.0 version.  The age respondent at marriage, age of spouse at the time of marriage, delivery at maternity hospitals, abortions and practicing any family planning method are exhibited negative regression coefficients, indicating that these are significantly and negatively associated with women fertility ordinal counts (p<0.05).  But, the residence, income, occupation, age of spouse, duration of marriage, spouse is employed, knowledge about contraceptives, satisfactory medical facilities available, opinion about number of children for comfortable life and spouse is a blood relative are significantly and positively associated with women fertility ordinal counts (p<0.05).  The ordinal regression model with complementary log-log built in link function has a maximum log likelihood value (-2019.8986) as compared to logit (-2989.1177), probit (-2078.3568) and Negative (n)log-log built link function (-3154.0832). Therefore, we conclude that the ordinal regression model with complementary log-log link function is better fit as compared to logit, negative (n) log-log and probit link functions in assessment of best contributing factors of women fertility.