Dealing with Threats to Internal Validity That Randomization Does not Rule Out

Abstract

Cook and Campbell (1979) identify five threats to internal validity that they claim cannot be controlled by using control group designs with random assignment of subjects. In this article, I suggest a design in which the control group subjects are given an alternate treatment that they perceive as equally desirable and that is similar in duration and procedures to the experimental treatment but that is concerned with dependent variables that are unrelated to the experimental treatment. I suggest that this design rules out or reduces all five internal validity threats, and I discuss the reasons supporting my opinion.