Off-campus UMass Amherst users: To download dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.
Non-UMass Amherst users, please click the view more button below to purchase a copy of this dissertation from Proquest.
(Some titles may also be available free of charge in our Open Access Dissertation Collection, so please check there first.)
An evaluation of the survey design used by the World Health Organization in epidemiologic research on AIDS in Africa
Abstract
A Monte Carlo model simulating actual populations was employed to evaluate precision in estimation of a standardized cluster sampling method proposed by the Global Programme on AIDS of the World Health Organization for general use in collecting population-based data on HIV seroprevalence. This study suggests that in real populations, where there is likely to be pocketing of infection, the proposed methodology will generally fail to provide precise estimates of the true population seroprevalence rate. In contrast, differences in precision of estimates of relative risk obtained using the modified EPI (Expanded Programme on Immunization) survey method versus simple random sampling (SRS), were generally of little practical significance. Given the added cost and difficulty encountered in implementing SRS at the second stage of sampling, the EPI-like methodology warrants further consideration for use in estimating relative risks. If differential levels of nonresponse are anticipated then it appears that the EPI-like method of sampling may spuriously provide parameter estimates with precision nearly equaling or exceeding that of SRS. If the primary objective of a particular survey is to construct confidence intervals that include the true population parameter, the EPI-like survey technique appears to be a reasonable sampling alternative, although this is likely to be accomplished at the expense of wider confidence limits and acceptance of less precise point estimates. This study also suggests that random selection of only one adult per household will improve the precision of resulting population estimates. However, since selection of only one adult per household would require that more households be visited, any gain in precision would need to be weighed against the likely increase in cost of household visitation.
Subject Area
Public health|Biostatistics
Recommended Citation
Harris, Donald Robert, "An evaluation of the survey design used by the World Health Organization in epidemiologic research on AIDS in Africa" (1990). Doctoral Dissertations Available from Proquest. AAI9110150.
https://scholarworks.umass.edu/dissertations/AAI9110150