Abstract
A taxometric model is developed for testing the existence of an underlying dichotomous taxonomy consisting of a 'taxon' and its complement. The model includes procedures for estimation of the base rate of the taxon and the valid positive and false positive rates for each indicator of the taxon. The model also includes tests (referred to as "consistency tests") for checking the internal validity and the goodness of fit of the model to the data. In an illustrative application, responses to interview questions were used as candidate indicators of a conjectured taxon of dementia in a sample of 841 New York City and London older people. In this application consistency tests indicated the existence of a taxon, since the fit of the model was extraordinarily good. Strong agreement of the taxometric classification with a clinical diagnosis of dementia based on a review of a case summary provided favorable evidence for the construct validity of the taxon.

This publication has 3 references indexed in Scilit: