Evaluating Technical Adequacy Features of Sentence Verification Technique as a General Outcome Measure of Content Knowledge

Abstract

Once students have mastered the mechanics of reading, they are expected to learn new material by reading. This new material, however, becomes increasingly more complex as students enter upper elementary and especially middle and high school. If students fail to comprehend what they read, they risk failure in content courses such as science and social studies. Early assessment of risk and appropriate response to that risk is a goal of effective education. One problem with the risk reduction sequence is that there are limited formative assessments that have been validated as technically adequate for assessing content knowledge. The present study examined an established reading comprehension assessment called sentence verification technique (SVT) as a formative measure of science content knowledge. SVT probes were administered to 130 fourth, fifth, and sixth grade students at 2 PK-6 schools for 5 months, as well as the abbreviated Stanford Achievement Test, 10th Edition, as a criterion measure. Monthly SVT probes were analyzed for internal scores consistency reliability, as well as for predictive and criterion validity. Multilevel modeling was used to determine if SVT was a significant predictor of student growth. Item types were examined to determine if there were significant differences in scores based on race or gender. Results indicated that SVT probes had internal consistency reliability estimates that ranged from .45 to .84, and criterion validity estimates ranged from .33 to .53. Sentence verification technique was found to have predictive validity for fifth and sixth grade, accounting for 24% to 40% of the variability in the criterion measure. Estimates for fourth and fifth grade showed that SVT was a significant indicator of growth. Finally, item analysis showed that there were marginally significant differences for gender and highly significant differences for race on items by type. SVT shows potential for use as a general outcome measure of content. While it has been shown to demonstrate internal reliability, predictive and criterion validity, and growth measurement capacity, more research is needed. Findings to date suggest that given the more complex nature of instruction in content, SVT may work best in combination with other validated general outcome measures, including those with academic language indicators such as vocabulary matching or critical content monitoring.