Search

This report is the first of a series investigating aspects of cognitive disorder in schizophrenia, and deals with the validity of some tests as measures of thought-process disorder ("formal thought disorder") and with the relations between the tests. A battery of tests, including the Object Classification, modified Proverbs, and Bannister-Fransella tests, was administered to a group of schizophrenics with active psychotic symptoms, selected so that there were equal numbers of acute and chronic, paranoid and non-paranoid, and male and female patients. A psychiatric rating of degree of thought-process disorder was made. The main results were: 1. The Bannister-Fransella measures of Intensity and Consistency are significantly and positively correlated among Acute Schizophrenics, but not among Chronics. 2. The Payne-Hewlett measures of "Non-A" responses and Proverbs are positively, but not significantly, correlated among both Acutes and Chronics. The present evidence supports the view that they should not be combined. 3. Seven of the eight correlations between Bannister-Fransella and Payne-Hewlett measures are negative, though non-significant. 4. Consistency is significantly related to a psychiatrist's rating of thought-process disorder in the expected direction among Acute, but not among Chronic, Schizophrenics. The measure has, therefore, been validated against this new criterion for Acute but not for Chronic Schizophrenics. The pattern is similar for Intensity, though the validity coefficient for acutes is not significant. 5. "Non-A" responses are heterogeneous, mostly unrelated to the usual meaning of "over-inclusion", and negatively related to the rating of thought-process disorder among Acutes and insignificantly among Chronics. The number of words used to define Proverbs is negatively, and almost significantly, related to rated thought-process disorder among Acutes. The positive relation among Chronics is small and non-significant.

In the biological and social sciences the research worker is frequently faced with a two-dimensional, entity by variable, matrix of numbers. The matrix may contain scores of persons on tests, judgments of persons or objects by persons, measurements of physical characteristics of insects, economic statistics of areas, party votes in electoral districts, demographic or epidemiological or ecological counts of areas, and so on.

The reviewer here outlines the theory behind Patrick Slater's technique for the analysis of Repertory Grids. He then presents an analysis of a very simple Grid. The clinical possibilities of the method are assessed, and it is put into its context as a research design by a comparison with other multivariate methods.

A group of males with XYY sex-chromosome complement was compared with a matched group of fellow-patients in a state mental hospital. There were very few significant differences between mean scores of the two groups on psychological tests. There were, however, differences in the inter-relations of tests. There was a marked discrepancy between the two groups in the effect of response set (defensiveness) on replies to questionnaires. A difference in the structure of aggression or hostility was also detected. If response set is indeed an important determinant of psychological test scores, and if its mode of action differs in XYY patients and in comparable non-XYY patients, then any obtained mean differences (or similarities) are suspect. The importance of this caveat becomes even more evident when we consider that the unco-operative patients who had to be omitted from the testing were probably high defenders. Perhaps the most direct approach to eliciting the psychological peculiarities of XYY patients would be to find out what it is that the high defenders among them do not wish to admit. The patient's and the psychologist's perceptions of the purpose of a particular test item are often quite disparate. The next stage of the enquiry is to ask what the patient thought he would be admitting if in fact he had not exercised defensiveness in a particular response.

In a recent article (Hope, 1969a) the author has demonstrated the theoretical relations between what psychologists call the analysis of persons and the analysis of tests. The demonstration was illustrated by an analysis of a hypothetical matrix M whose rows were called entities and whose columns were called variables. It was shown that, by combining Slater's (1964) method of spherical co-ordinate mapping with a method of numerical taxonomy suggested by the author, it is possible to perform a comprehensive and self-consistent pair of analyses whose conclusions can be read off three diagrams: a map of variables, a dendrogram of entities, and a profile diagram of the multivariate means of groups of entities.

A psychotic v. neurotic scale has been presented in the manual of the Symptom-Sign Inventory (Foulds and Hope, 1968). This scale was constructed after comparing each of four female neurotic groups with each of four female psychotic groups and each of four male neurotic groups with each of four male psychotic groups. For an item to be included in the scale 14 of the 16 comparisons within each sex had to show differences significant at least at the 5 per cent. level. On 18 items the psychotic groups had frequencies in excess of the neurotic; on only one item was it the other way round. The scale score is the sum of these 18 items minus the one.