Classification Accuracy of Curriculum

We examined the classification accuracy of Curriculum-Based Measurement in writing (CBM-W) Picture Word prompts scored for words written (WW), words spelled correctly (WSC), and correct word sequences (CWS). First graders (n = 133) were administered CBM-W prompts and the Test of Written Language–Third Edition (TOWL-3; Hammill & Larsen, 1996). Prompts scored for WSC showed acceptable levels of sensitivity (.947) and specificity (.587) with the TOWL-3 Contextual Language. Positive predictive values were low (approximately .20 to .30), and negative predictive values were high (mostly above .95). Overall classification accuracy, represented by the area under curve (AUC), ranged from .727 to .831. Further research regarding ways to improve classification accuracy of CBM-W and preliminary implications for practice are discussed