A Simple, Exact Decomposition of Selective Recruitment and Drop-Out Effects in Longitudinal Data

Evolutionary biologists have developed very interesting methods for decomposing change in a variable measured at the population level.  As far as I know, these methods are not known in the Social Sciences.  If you have panel (cross-sectional) data with individuals uniquely identified, and linked if they appear in more than one panel, it is possible to decompose change between pairs of panels at the population level into three exact and additive components: change measured for the ‘stayers’ who appear in two or more consecutive panels; change created by selective recruitment; and change created by selective attrition (drop-out). The decomposition is fast and exact, and not statistical, and can be applied to both means and additive measures of variance.

• Register here