This paper considers the possible effects of making inferences about individuals from aggregate data. It assumes a knowledge of regression analysis, and explores the utility of techniques designed to make the inferences in causal modelling more reliable, including a comparison between ecological regression models and ecological correlation.