(48, 46, and 47 percent). It seems unlikely that the rate of brain dysfunction would be the same in neurological patients, dull-normal, and learning disabled; indeed it seems likely that brain disorder would be more common in the neurological patients than in the other two abnormal groups. But if groups differ in the base rate of some condition, then a valid test for that condition must show different rates in these groups (2). In other words, the constant proportion of abnormal EEG's in the three different abnormal groups leaves us with two possibilities: either the rate of brain disorder is equal in these three groups or the EEG test has no validity for detecting brain disorder.It might be argued that the data do at least support the validity of the EEG test for distinguishing normal from abnormal children, where abnormal now designates a behavioral category broad enough to comprehend neurological patients, dull-normals, and the learning disabled. The meaning and usefulness of this new combined category remain to be established.