Psychometric Characteristics of the Behavioral Observation Scale
Abstract
Self-, peer-, and supervisor ratings were obtained on 52 psychiatric aides using a Behavioral Observation Scale (BOS). Self ratings showed less leniency error than peer- and supervisor ratings. Halo error could not be assessed due to a negative correlation between means and variances. A multitrait-multimethod (MTMM) analysis supported the presence of strong rater bias and significant convergent validity but not discriminant validity. The results of the analyses demonstrated that the ratings obtained from a BOS were not psychometrically superior to other appraisal formats. Questions were raised as to the adequacy of a five point scale, data transformation, and rating scales.
This paper has been withdrawn.