Measures of Writing Skills as Predictors of High Stakes Assessments for Secondary Students

Abstract

This study examined the potential utility of written expression scoring measures, developed in the curriculum-based measurement research, to monitor student progress and predict performance on a high stakes state mandated assessment for high school students. In response to a teacher generated prompt, 10th-grade students completed 3 brief constructed response (BCR) and 2 extended constructed response (ECR) writing samples throughout the academic year. Writing samples were scored for total words written (TWW), words spelled correctly (WSC), correct writing sequences (CWS), correct minus incorrect writing sequences (CMIWS), percentage of words spelled correctly (%WSC), percentage of correct writing sequences (%CWS), production dependent index, and production independent index. The average time to score a BCR for TWW, WSC, CWS, and CMIWS was over 7 minutes, and the average time to score an ECR was over 16 minutes. Alternate form reliability correlation coefficients between scoring measures were only in the weak to moderate range. Results revealed that girls wrote more words, spelled more words correctly, produced more correct writing sequences, and produced more correct minus incorrect writing sequences. Across writing samples, statistically significant but small increases were found on scoring measures. Results of multiple regression and logistic regression analyses failed to provide a model that accurately predicted student outcomes.