21 - The Potential-Outcomes Model of Causation

Summary

In this final part of the book I explore one of the most influential theoretical perspectives on causal inference in the social sciences. Chapter 21 outlines the model of causality, in which the effects of an experimental or observational treatment are defined in terms of the potential outcomes that could have occurred under different possible interventions. Chapter 22 provides a classic example from the causal inference literature, highlighting what we gain by thinking within the potential-outcomes perspective. Chapter 23 provides an example of the impact of migration on social solidarity using survey data from Mexico.

Let's start by discussing what causal inference is. One goal of using quantitative data and techniques is to model the relationship between an outcome and a series of variables that predict and answer specific questions about a phenomenon of interest. Probably the basic purpose of quantitative research is to identify causal relationships between outcomes and their predictors. A causal relationship occurs when one variable, under certain contextual conditions, increases the probability that an effect, manifested in another variable, will occur (Eells 1991; Gerring 2005). We can never be sure, though, that X caused Y (especially when we use observational rather than experimental data) because there are so many possible confounding variables that we cannot control for, every single one of which might have a direct or an indirect effect on Y.