THE DIAGNOSTIC ACCURACY OF THREE COMPUTER-ADAPTIVE SCREENING MEASURES OF READING

Abstract

The data from universal screeners must be valid and reliable in order to use it to make appropriate decisions about how best to allocate resources to support students who are at risk of not passing the state achievement test. The instruments used as part of universal screening must also have diagnostic accuracy. This study examined the diagnostic accuracy of three computer-adaptive universal screening measures of reading for predicting student performance on the state achievement test. Using extant data, spring universal screener scores from second graders in public schools in a southeastern state were used to predict their performance on the state achievement test administered in the spring of their third-grade year. Logistic regression and receiver operating characteristic curves were used to examine the diagnostic accuracy of Istation Indicators of Progress Early Reading, MAP-Reading, and Star Reading. Results show that all three computer-adaptive universal screeners were strong predictors of the ACT Aspire, an assessment used as the statewide achievement test. However, the Istation and MAP universal screeners were reliably more accurate than the Star Reading assessment based on area under the curve comparisons. In analyses of based on cut scores associated with various means of standard setting, none of the screeners met the recommended levels of .90 sensitivity and .70 for specificity. Implications for use of the computer-adaptive screeners in schools as part of a universal screening process are discussed.