An Investigation of the Validity of Scores on Locally Developed Performance Measures in a School Assessment Program

Abstract

A multitrait-multimethod approach was used to investigate the validity of scores on locally prepared performance assessments. The multiple traits were reading, language arts, and mathematics, and the multiple methods were the Comprehensive Test of Basic Skills/4, locally developed multiple-choice tests, and locally prepared performance assessments. The results offer very limited evidence of convergence and discrimination for scores on the performance assessments. A partial explanation for the lack of construct validity evidence for the performance assessment scores is the probable low reliability of scores on the performance measures due largely to their limited content sampling and relatively low objectivity of scoring.