ABSTRACT
Internal validity is critical to the defensibility of impact evaluations. External validity, statistical validity, and measurement reliability and validity pertain to all program evaluations, no matter whether they are primarily descriptive or causal. In contrast, internal validity pertains only to impact or causal evaluations, and it is key to their credibility, because it is defined as the accuracy of the causal claim. If an evaluator claims that program X (e.g., welfare reform) causes outcome Y (e.g., reduced welfare caseloads), and if that claim is accurate, then it is said to be an internally valid claim. Similarly, if an evaluator claims that program X (e.g., legal political action committee (PAC) contributions to members of the U.S. Congress) does not cause outcome Y (e.g., how mem bers vote on bills relevant to the PAC), and if that claim of no impact is accurate, then the claim is said to be internally valid.1
