Conditional probabilities and the grid test for schizophrenic thought disorder

Abstract

The grid test for schizophrenic thought disorder was administered to a sample of 149 psychiatric patients. This sample was drawn from a population with a known base rate of schizophrenia and an estimated base rate of thought disorder, thus allowing for the calculation of population specific cutting scores for the grid test.for diagnostic efficiency using conditional probability formulae. a diagnostic instrument.The grid test, using these data along with others from previous grid test studies, was then analysedThe conclusion reached is that the grid test for schizophrenic thought disorder is unsatisfactory asBannister & Fransella reported the development of the grid test for schizophrenic thought disorder (GT) in 1966 and published it in standardized form in 1967. The GT yields two scores: intensity (I) and consistency (C). Bannister & Fransella found, in setting cut-off points a t I < 1000 and C < +0.49, that 80 per cent of thought-disordered schizophrenics (TDS) and only 6.4 per cent of non-TDS, other psychiatric patients and normals in their sample scored below these points. Ninety point seven per cent of the subjects in their sample were correctly classified by the GT a s either T D S o r other than TDS.contexts. A cursory review of the literature from 1966 through 1978 indicates that at least 31 papers o n the topic of the GT have been published. Five of these papers, though not all specifically designed as replication studies, report data in a form suitable for comparison to Bannister & Fransella's (1966) original report. The findings of these studies are summarized in Table 1 . As is obvious from the summary, no one t o date, including the GT authors, has been able t o replicate either the valid and false positive rates o r the overall correct classification rate reported in the original GT study.Following its publication, investigators went on t o study the GT in a variety of