Modelling the Impact of Poverty on Contraceptive Choices in Indian States: A Multilevel Analysis
Isabel Tiago Oliveira, Sabu S. Padmadas, José G. Dias
Department of Social Research Methods, ISCTE-Lisbon University Institute, Lisbon, Portugal; Centre for Global Health, Population, Poverty & Policy, University of Southampton, Southampton, Portugal; Department of Quantitative Methods & UNIDE, ISCTE-Lisbon University Institute, Lisbon, United Kingdom

The underlying relationship between poverty and contraceptive choices is mediated by individual and household level factors. In India, 56.3% of current married women use contraception, mostly by means of female sterilization. The significance of female sterilization in contraception is outstanding; however, it is associated with a stopping behavior strategy. On the other hand, modern and traditional contraceptive methods allow women, not only to control the number of children, but also adjust the spacing between births, with major consequences in the mother and children health. The choice between female sterilization (FS) and modern spacing methods (MSM) are associated with women socio-economic situation.

Both the wealth index and female education are associated with the women contraceptive choices. For the Indian married women currently using contraception, around 70% prefer FS in the four first quintiles, but in the highest one this option decreases to half of the women using any contraceptive method, simultaneously the use of MSM increases with wealth, particularly in the 4st and 5st quintiles. As for educational level, the gradient is even stronger than household wealth. FS is the most important contraceptive method in the lower educational levels, and its relative importance decreases as educational level increases. On the other hand, the percent of women using MSM, within all married women using contraception, rises markedly with educational level. The most educated women prefer MSM over FS – a unique situation in Indian context.

Using data from the 2005-06 National Family Health Surveys, this research systematically examines the poverty impact on differential contraceptive female options, in Indian states. Latent class models are used to classify households by their wealth status. The interaction between household wealth and other factors in determining the contraceptive choices of women is modeled using multilevel multinomial logistic regression models, controlling for relevant individual, household and community level factors.

Keywords: Latent class; Multilevel multinomial logistic regression; Wealth; Contraception

Biography: Demographer. Work usually in fertility and mortality.