Causal Analysis in Population Studies

Zusatztext

<P>The central aim of many studies in population research and demography is to explain cause-effect relationships among variables or events. For decades, population scientists have concentrated their efforts on estimating the `causes of effects¿ by applying standard cross-sectional and dynamic regression techniques, with regression coefficients routinely being understood as estimates of causal effects. The standard approach to infer the `effects of causes¿ in natural sciences and in psychology is to conduct randomized experiments. In population studies, experimental designs are generally infeasible.</P> <P>In population studies, most research is based on non-experimental designs (observational or survey designs) and rarely on quasi experiments or natural experiments. Using non-experimental designs to infer causal relationships¿i.e. relationships that can ultimately inform policies or interventions¿is a complex undertaking. Specifically, treatment effects can be inferred from non-experimental data with a counterfactual approach. In this counterfactual perspective, causal effects are defined as the difference between the potential outcome irrespective of whether or not an individual had received a certain treatment (or experienced a certain cause). The counterfactual approach to estimate effects of causes from quasi-experimental data or from observational studies was first proposed by Rubin in 1974 and further developed by James Heckman and others.</P> <P>This book presents both theoretical contributions and empirical applications of the counterfactual approach to causal inference.</P>

Weitere Details