Current nosological thinking, as exemplified by the American Psychiatric Association's Diagnostic and Statistical Manual (l*), makes use of terms such as anxiety, reality relationships, and concept formation as attributes distinguishing between diagnostic groups. Studies by Ash('), Doering('), Mehlman(l1) and recently Schmidt and Fonda (13) have demonstrated that diagnostic categorization may introduce considerable rater error. However, diagnostic categorization is readily obtainable information and offers the research worker a criterion for studies of the variables which conceptually differentiate the diagnostic groups. Possible difficulties in this procedure were brought into focus by Zimet and Brackbill(16). In their study, scores of four common and relatively well validated test measures of anxiety were examined for 97 patients tested at a VA General Medical Hospital whose discharge diagnosis had been neurosis, psychosis, or personality disorder. The only differentiation between diagnostic groups by the test measures of anxiety was that on the Welch Anxiety Index(16) neurotics saw themselves as less anxious than the other two P. 176).The importance for research of the finding that the frequently used criteria of tests and diagnostic categorization are not consistent led to the present study. This investigation (a) repeated Zimet and Brackbill's study using patients admitted to a neuropsychiatric hospital (and therefore presumably more severely ill) ; (b) gathered data of additional test measures of variables of mental functioning used in nosological conceptions; and, (c) compared three test measures of anxiety on groups of patients given the same diagnosis a t two diflerent hospitals.
METHODEvery male patient entering a VA neuropsychiatric hospital admission ward during three months, who was not brain injured or over 50 years of age, was administered the research test battery within two weeks of admission. Eighteen months after testing, the established clinical diagnosis was determined from the discharge summary, or if the patient remained in the hospital, from the progress notes. The sample of 90 patients differed from Zimet and Brackbill's sample in that the percentage of psychotics was far greater in the present study, the percentage of personality disorders far lower. A second difference between the studies was that Zimet and Brackbill used records of patients obtained by referral and might have obtained a sample of difficult diagnostic problems. The present sample of all patients entering the admission ward during three months may be assumed to have included the easy diagnostic decisions as well as the difficult ones. A third difference between this investigation and that of Zimet and Brackbill was that in this study all data were collected by the same examiner while in the previous study tests were administered by a number of different examiners. The present study eliminated the possibility that the lack of differentiations between the groups reported by Zimet and Brackbill was a result of variation i...