Single-subject designs are tools for behavioral researchers to evaluate the impact of an intervention on an individual. Single-subject designs have four essential components: the taking of repeated measurements, a baseline phase (A), a treatment phase (B), and data analysis (often based on graphing).
Single-subject A–B designs can be improved by adding a treatment withdrawal phase (A–B–A), by using multiple subjects at different baselines, or by trying multiple treatments. Repeated measurement controls for many of the potential threats to internal validity. The period between the last baseline measure and the first treatment measure is susceptible to the effect of history.
Results of single-subject designs can be difficult to interpret due to discrepancies in baseline measures, delayed changes, improvements in the baseline period, and distortions due to graphing. Some researchers use statistical tests to lessen these problems. Generalizability from single-subject designs entails direct replication, systematic replication, and clinical replication.