| Outcomes on summative measures of achievement such as end-of-course assessments, which are often referred to as "high-stakes tests" because the results determine whether students have successfully met a graduation requirement, are likely the product of multiple factors. Students' aptitude, instructional practices and teacher attributes, characteristics of the assessment instrument, and other factors contribute to students' scores. Because only instructional factors are usually open to change by teachers, it should be valuable for teachers to have a trustworthy means of predicting students' outcomes as their knowledge and skill is forming during the course of instruction, thus permitting the teachers to make adjustments in instruction that will increase the chances of students passing the high-stakes test. One method of assessing students' competence formatively is to collect data during the course of study, and one of the most extensively studied systems for collecting such formative data is often referred to as "curriculum-based measurement" (CBM). Researchers who have examined the psychometric characteristics of CBM measures have found that CBM data collected during the course of study in the elementary areas of reading and arithmetic predict students' outcomes on norm-referenced and high-stakes tests Crawford, Tindal, & Steiber, 2001; Fewster & MacMillan, 2002; Helwig, Anderson, & Tindal, 2002; McGlinchey & Hixson, 2004; Sibley, Biwer, & Hesch, 2001). Although the technology exists for using CBM techniques in high-school content-area subject (sciences and social studies), no studies have examined whether these formative measures reliably predict students' outcomes on high-stakes tests. This study examined the relationship between scores on CBM measures, in the form of vocabulary-matching mini-tests and high-stakes summative assessments in social studies.; Participants were three high school U.S. history teachers and their students. Students (total n=69) were in advanced placement (n=20) honors (n=27), and general level (n=22) classes. Thirty-nine students were female, and 30 were male. Independent variables included: (a) six researcher-developed CBM probes, or mini-quizzes, (b) a measure of general vocabulary, (c) class assignment, (d) class rank as determined by grades in the previous grading period, and (e) gender. Dependent variables were (a) a researcher-developed outcome measure based on Virginia's high-stakes assessment and (b) the Virginia high stakes Standards of Learning (SOL) test.; The correlations between scores on CBM probes and both the researcher-made outcome measure (r=.327) and the Virginia SOL (r=.601) were statistically significant at the 0.01 level. Only class assignment and general vocabulary contributed to the prediction of students' scores on the mock outcome measure. However, for the actual high-stakes SOL assessment, class assignment, CBM score, and rank in the class contributed significantly to the prediction of outcomes. These findings indicate that formative assessment, in the form of CBM probes, is a good predictor of scores on high-stakes tests. This information can help teachers predict early in the semester which students are prepared and which students are not prepared for high-stakes tests. Given the capacity to predict outcomes, teachers can alter instruction and increase achievement. |
