Demographic measurement and analysis in developing countries: Three essays

This dissertation is composed of three essays examining aspects and issues of demographic measurement and analysis in developing countries. The first essay uses reinterview data collected in 2001 by a household survey conducted in rural Malawi to examine individual consistency in response to questions about HIV/AIDS and other topics. The nature and implications of individual inconsistency are further analyzed by evaluating covariates of individual consistency and the implications of inconsistency for univariate and multivariate estimates. The main finding is that the reinterviewed respondents are overall consistent in their answers and that, when there are inconsistencies, they do not significantly affect the conclusions that can be drawn from multivariate analyses of the survey data. The second essay uses couple data collected by the same household survey to evaluate husband-wife agreement in reporting over time. Couple discrepancies in reporting are analyzed from a methodological perspective, by developing models that consider them as an indicator of response error arising because of specific gendered interactions. The main finding is that couple discrepancies in reporting arise because of systematic bias in husbands\u27 and wives\u27 answers to the questions considered. I also find that, if not accounted for, this bias might have serious implications for panel analyses of these data. The third essay discusses the relevance of biases in sentinel surveillance data for projections of HIV/AIDS in China. It is currently widely acknowledged that existing projections are limited by the availability of scientifically valid data. However, it is less emphasized that additional biases are introduced by shortcomings in the projection methodology, which does not account for the heterogeneity of HIV rates and risk in different Chinese provinces. In this essay I focus on the latter source of bias for projections of HIV/AIDS in China and I show that the existing knowledge about the epidemic of Yunnan province (which accounted for about 40 per cent of China\u27s HIV-infected population at the end of 2000) gives important insights about the possible evolution of HIV/AIDS in other provinces as well as nationwide