The Reliability and Construct Validity of a Behaviorally Anchored Rating Scale Used To Measure Nursing Assistant Performance

Data are presented that demonstrate the reliability and construct validity of a 27-item Behaviorally Anchored Rating Scale (BARS) used to rate nursing assistant performance. A sample of 757 nursing assistants from 44 randomly selected long-term care facilities were rated with the BARS instrument by a team of independent observers, and later by the nursing assistants'supervisors. Interrater reliability between the independent observers and supervisors ranged from 0.34 to 0.96 and averaged 0.78. The internal consistency reliability measured by Cronbach's alpha was 0.71 for the independent observers, 0.76 for the supervisors. Construct validity was assessed by exploratory factor analysis. The underlying factor structure had 3 factors that accounted for 84.0% of the variance. Results were interpreted as supporting the reliability and indicating construct validity of this BARS for nursing assistants. Because BARSs have been shown to reduce subjectivity and personal bias, they are recommended for evaluating nursing assistant performance.