Accounting for areal and temporal differences using decomposition procedures: an illustration

Abstract

This paper illustrates in details how to use decomposition procedures to account for areal and temporal differences in the level of mortality, using the case of infant mortality in Kenya. Logistic regression is used to decompose the effects of various factors on the risk of mortality. The major advantage of the methodology described here is that it can identify the factors that account for differences in mortality levels between two or more places, and, if used or temporal mortality changes in the same place. In essence, these methods if used correctly, disentangle differences in the values of explanatory variables of mortality between two regions, or between two different time periods in the same place that are due to the differences in the values of the explanatory variables , and those that are due to the structure of relations between mortality and the explanatory variables.