Selecting criteria for the evaluation of secondary school assistant principals based upon item discrimination power
This study was conducted for the purpose of identifying discriminating, reliable, and valid criteria for evaluating secondary school assistant principals which could be used by local schools in constructing a new or refining an existing evaluation instrument. Data were collected by having 805 principals, assistant principals, superintendents/central office supervisors and teachers complete a 50-item questionnaire to determine which criteria had the ability to separate high assistant principal performance from that of average and below average performance;Forty-four assistant principals from 16 school districts located in Iowa, Kansas, Texas, North Carolina, Indiana, and Michigan volunteered to participate in this research effort. Each of these assistant principals were evaluated by 15 or more raters;Major findings of this study include: (1) all 50 items on this questionnaire discriminated or measured differences between the job performance of the assistant principals being rated at the.05 level of significance, (2) 25 items on the instrument discriminated at the.01 level of significance, (3) 2 or more evaluator groups had significantly different means for 30 of the 50 items, (4) assistant principals were ranked significantly lower by principals than the other rater groups on 16 of the items, (5) central office supervisors/superintendents ranked assistant principals higher than the principal rating groups on 14 items, (6) the rater groups most frequently agreeing with one another were the teacher rating group and the superintendent/central office supervisor rating group;It is recommended that items being used for assistant principals' performance evaluations be selected carefully so that evaluating items are matched with the particular job responsibility for the assistant principal. It is further recommended that as many persons as possible be involved in the evaluation process so as to offset personal bias.