MATHEMATICS PERFORMANCE ASSESSMENT - TECHNICAL QUALITY AND DIVERSE STUDENT IMPACT

Reform in mathematics education calls for a closer tie between instruc tion and assessment. This study developed performance assessments from hands-on instructional activities and examined their reliability and validity for obtaining individual achievement data in large-scale surv eys. The major source of unreliability was the tasks, not the raters. Many tasks are needed to get a dependable measure of a student's mathe matics achievement. With respect to validity, results suggested that t he performance assessments measure different aspects of mathematics ac hievement than do traditional multiple-choice tests. Moreover, the per formance assessments, but not the multiple-choice test, distinguished the performance of students in hands-on and traditional curricula with the former scoring higher, on average, than the latter. Ethnic group comparisons indicated that Anglos scored higher, on average, than Lati nos on all achievement measures. The magnitude of the difference varie d by the curricular experience of the student. For students in traditi onal curricula, qualitative analysis indicates that Anglo and Latino s tudents approached the problems similarly, made the same types of erro rs, and employed the same strategies in solving the mathematics proble ms.