validity of peer ratings in the selection of potential school administrators

Florida State Law (1981) requires each school district to implement a researched-based method for identifying and selecting school administrators. In 1986 the Palm Beach County, Florida school district began a Potential Administrator Program. Legal and utilitarian concerns required that analyses of various instruments and methods of candidate assessment and selection be undertaken. The purpose of the present study was to examine the relative validity of peer raters compared with other sources in selecting participants for inclusion in a principal training program. The other sources were superordinate and self-ratings. The relationship between peer raters who were nominated and those who were not nominated by the candidates was also explored to determine the effect of using nominated raters exclusively. The sample for this study was drawn from 262 secondary school teachers in Palm Beach County who were eligible to apply for the training program. Using a Likert type scale, peer raters (n = 958) and superordinate raters (n = 42) evaluated the subjects (n = 54) on thirteen behavioral dimensions determined by research to be characteristic of high performing principals. Each candidate also completed a self-rating on the same behavioral dimensions. Finally, each candidate underwent a work sample assessment based on three of the dimensions. This assessment was divided into a communications and a content portion for evaluation purposes. Thirty candidates were chosen for the training program. An analysis of multi-trait multi-method matrices of the rating sources and behavioral traits, as well as an analysis of variance of the composite rating source means, and a stepwise multiple regression analysis predicting the work sample assessment from the three ratings sources indicated support for using peer ratings in selecting principal candidates. Peers added unique information to that provided by superordinate and self-raters relative to the targeted dimensions. Data on using raters nominated by candidates was inconclusive. Analyses were performed on: (1) a multi-trait multi-method matrix of nominated and non-nominated peer ratings and behavioral traits; 2) an analysis of correlations between these two rating sources and superordinate ratings, self-ratings, and work sample assessments. Further studies using larger samples and including elementary school candidates are recommended.