Dealing with missing data in demographic and health surveys

Abstract

This project deals with the problem of missing data in the Tuunganepillars of Hope -PWP (2008) survey carried out by impact Research and Development organization (IMPACT RDO)in Kisumu, Kenya. The emphasis of this project ison Expectation Maximization (EM) algorithm technique as a method of handlingdata problem in multivariate analysis. The primary objective of the 2008 PWP project was specifically to enhance supportive and informed home environments of HIV-infected youths and promote safe sexual behaviors and generallyto provide up to date information for policymakers, planners, researchersand program managers, which would allow guidance in the planning, implementation, monitoring and evaluation of health programs in Kenya. However,when faced with missing values, researchers or practitioners in research usuallyproceeds as if the omitted cases had never really been observed.