Researchers have long been aware that "pretest sensitization" is a potential threat to the external validity of experimental studies. Only recently, however, has it been suggested that "posttest sensitization" might also limit the generalizability of results. This latter phenomenon would exist to the extent that the effects of a treatment are latent and appear only when a posttest is administered. Under this condition, the treatment would work differently for subjects in the researcher's sample than for other individuals who might later receive only the treatment without the accompanying posttest. To be scientifically useful, it is imperative that the hypothesis of posttest sensitization be capable of empirical verification, and a quasi-experimental design is proposed for this purpose. (Author)