The primary purpose of this dissertation is to assess the extent to which household characteristics and behaviours exert their effects both directly and indirectly on childhood mortality through the more proximate factors that can be measured within the context of society. A child mortality model, primarily based on Mosley and Chen's framework, is developed by linking individual and societal factors. Then the model is tested with empirical data from the Fars Province of Iran. The survey data were collected in 1991--92 in five counties of Fars. It consisted of 10665 interviews and covered 67 villages 14 towns and one city. Three sampling techniques were employed: (1) proportional stratified sampling; (2) cluster sampling; and (3) simple random sampling. / Three levels of analysis were carried out in this thesis: individual, societal and contextual. Bivariate and multivariate logistic regression analysis were done for subsets of variables based on the child mortality model that were identified to be good predictors of child mortality and which were also identified theoretically as proximate and intervening variables. / The individual level analysis reveals that place of residence, education of the mother, and occupation of the father from the socio-economic, factors; and age of marriage of mothers, pregnancy order, and pregnancy age from demographic factors; and visiting doctors during pregnancy, type of delivery, pregnancy duration, birth weight, and vaccination from the health status factors; and housing quality are the important determinants of child mortality in Fars. / At the societal level, rural setting, the literacy rate of the villages and assets indexed by sheep per capita are the important determinants of child mortality. Also child mortality rate differentials were found to be compatible with that of additive developmental index of regions (counties). / Contextual analysis shows that birth weight, pregnancy duration, pregnancy order, and house facilities are, in Iran, significant predictors of child mortality. Among all the variables, these variables appear to be the most proximate variables and the other variables, including socio-economic and demographic variables, significant intervening variables. / The results of this dissertation support the claim that child mortality can be a sensitive indicator of human development and quality of life both at the individual and societal levels. Most significantly it appears to be prerequisite to fertility decline. The most important finding from these analyses is that child mortality is influenced both by the individual's characteristics as well as by community characteristics. In better words, social organisation as proposed in the child mortality model matters.

No description available.