The problem with categorical thinking by psychologists.

Continua abound in the natural world, but the treatment of these continua in the psychological study of behavior is often categorical. Four practices and their consequences are considered: limited sampling of a continuum, turning continuous predictors into categories through dichotomization and similar practices, treating continuous predictors as categorical in an analysis, and overuse of the analysis of variance. Concrete examples illustrate the consequences of these practices. Recommendations to improve the treatment of continua, both in the design and analysis of data, are provided.