Abstract

With the institution of the No Child Left Behind Act (NCLB, 2001), America's public education has dramatically shifted to a system driven by accountability and assessment. This emphasis on assessment has focused increased attention on the strategies and methodologies with which we measure student performance and evaluate system effectiveness.

The focus of this study was to examine the relationship between the assessment paradigms used to measure student and system performance. Specifically it was the purpose of this study to examine two common testing models: standardized achievement tests and content based criterion-referenced tests. This study will examine the predictive strength of content specific criterion-referenced assessments to predict student performance on standardized achievement tests. The analysis will examine this relationship for a sample of students on a year by year basis, and the cumulative effect over a three year period.

Descriptive and inferential statistical analysis was performed on random samples of matched data for students in the 8^th, 9^th and 10^th grade levels. A multivariate canonical correlation statistical procedure was employed on student performance data in each sample group. This procedure yielded analysis of both the overall predictive strength of each combined model and the contributions of individual variables in accounting for the predictive relationship.

Analysis of the data showed significant predictive strength in each of the eight sample sets tested. Statistical analysis of each variable also demonstrated high levels of effect size for most all variables in the eight models. The levels of significance reported were such that all null hypotheses were rejected and it was concluded that a significant predictive relationship exists between student performance on criterion-referenced/End of Course assessments and performance on the Idaho Standards Achievement Test.