Screening in the Upper Elementary Grades: Identifying Fourth-Grade Students At-Risk for Failing the State Reading Assessment

This study examined the predictive validity and classification accuracy of individual- and group-administered screening measures relative to student performance on a year-end state reading assessment in two states. A sample of 321 students was assessed in the areas of word-level and text fluency, as well as reading comprehension in the fall of fourth grade. For individual measures, a group-administered reading comprehension test exhibited the highest classification accuracy (74%-80%) for both outcomes though no screener demonstrated optimal sensitivity and specificity levels. Using a multivariate approach, logistic regression results revealed minimal to no increase in classification accuracy over the single comprehension measure. Receiver operating characteristic (ROC) curve analyses determined local cut scores to maintain sensitivity constant at 0.90, which resulted in a large number of false positives. The implications and limitations of these findings for screening at the upper elementary level are discussed.