APPLYING THE JACKKNIFE METHOD TO SIGNIFICANCE TESTS OF DIAGNOSTIC DIVERSITY

The sample theory of normal diversity indices is complex. Distribution free methods, such as the jackknife method, can easily be used to dete rmine confidence intervals and testing diversity. Jackknife estimates and their variances for a number of different diversity indices are de scribed in this paper. A simple numerical example is given for demonst rating this method. Discrimination based on confidence intervals is al so discussed. It is assumed that there is a special correlation betwee n the sensitivity parameter m and the relative width of confidence int ervals in the Hurlbert index family. It is shown that the usual estima tion of the Hurlbert index coincides with the relating jackknife estim ate. For demonstration, diagnoses registered in a set of death certifi cates are used. There is a considerable diversity in diagnoses among d ifferent diagnostic groups: the diversity is largest in autopsy report s, whereas it is non-significant in GP's reports and in reports of phy sicians authorized to issue death certificates. Knowing that autopsy r eports tend to be fairly accurate, our research findings seem to confi rm the hypothesis that there is a correlation between reliability and diversity of diagnoses.