Statistical regression is a threat to external validity of experimental research, which uses subjects selected for participation based on extreme scores.
