Relationships Among Predictors in Longitudinal Data: Temporal-Sequential Analysis by Regression - TSAR

Abstract

In analyses of longitudinal data attention is appropriately drawn to criterion measures of growth. However, it seems appropriate to consider predictor variables, and to do so in a fashion which helps understand their interrelationships. A method of arranging predictors is described which draws on regression analysis, and so uses powerful inferential tests of statistical significance. Examples are given of patterns of predictors educed by representative analyses. Data are drawn from two data sets with attention to several measures in the first 5-6 years of life of a particular cohort, and to demographic data on childhood from several countries of the world, in the second instance.