Inter-rater reliability of PSE-9 (full version): An Italian study

The Research Unit of the Servizio di Psicologia Medica at the Institute of Psychiatry of the University of Verona, 1980–1991

“…The Italian text was then adjusted, where necessary, according to his comments. Inter-rater reliability of both instruments (Italian versions) were tested in our Unit (Mignolli et al 1988;Ardoin et al 1991) and found satisfactory. Patients living in the community were found to be in a variety of living conditions, but none who were judged to need support were without it (i.e.…”

Estimating psychiatric morbidity in the community Standardization of the Italian versions of GHQ and CIS

“…Finally a three way analysis of variance without replication was carried out, using as dimensions raters, subjects and symptoms, according to Maxwell and Pilliner (1968); we assumed that raters and symptoms had fixed effects and subjects had random effects, as Mignolli et al 1988) also did. In the analysis, the mean square for the main effect of raters was tested against the first order interaction between raters and subjects; the mean square for the main effect of items was tested against the first order interaction between items and subjects; the mean square for the main effect of subjects and items were tested against the three factor interaction mean square.…”

II. Impairments in Schizophrenics in Cross-sectional and Longitudinal Perspective — The Mannheim Experience in Two Independent Samples

“…Also, the use of indices of reliability is quite variable, as reviewed by Anker (1983), who suggested using at least two different coefficients simultaneously in order to avoid drawbacks and flaws inherent in all available methods to date. PSE behavioural items (on which part of the PIRS is based) had earlier been found more problematic than symptoms (Wing et al, 1974), but in a recent study (Mignolli et al, 1988) using well-trained interviewers mean kappa values of 0.95 are given for the PSE behaviour sections. Whereas some centres, using up to 14 different raters with only a brief training and few patients reported lower reliability figures with the PIRS as opposed to the PSE (Slooff, unpublished), we calculated a kappa of 0.79 and a pairwise agreement rate of 89.4 % on the basis of 58 interviews (after an initial training of at least 15 video and live PSE and PIRS interviews beforehand).…”