Ghiselli (1966) compiled the results of most of the published and unpublished studies on the validity of tests used to predict performance in different jobs. He found that observed validity coefficients vary considerably across studies even when jobs and tests appear to be very similar. Because of Ghiselli's work, most other psychologists concluded that a validation study must be conducted when the test is to be used in a different job, situation, company or group of applicants. This conclusion has been labelled the situational specificity hypothesis. This hypothesis holds that, if the setting does not vary, validity will not vary. Two studies were conducted to prove this prediction, the first study using a composite test and the second study using a psychomotor test. It was found that (a) even when the setting was held constant, there was great variability in observed validity coefficients; (b) the variability was similar for both composite and psychomotor tests. These findings contradict the situational specificity hypothesis. However, they are consistent with the predictions of the validity generalization hypothesis. Furthermore, the findings are consistent with results obtained by Schmidt & Hunter (1984).