An Investigation of the Reliability and Validity of Curriculum-based Measurement Maze Probes: A Comparison of 1-minute, 2-minute, and 3-minute Time Frames

Abstract

Prevention science has suggested that universal screening can enhance educational and mental health outcomes in the schools (Greenberg et al., 2003). A three-tier model of prevention has been proposed by Albers, Glover, and Kratochwill (2007) and Brown-Chidsey and Steege (2005) employing universal screening assessments of basic academic skills at Tier-1. Curriculum-based measurement maze (CBM-maze) probes are universal screeners that were developed as measures of reading comprehension. They are characterized as easy to administer, time-efficient, valid, and reliable (Parker, Hasbrouck, & Tindal, 1992). CBM-maze probes are short stories consisting of 400 words where every seventh word is omitted and replaced with three answer choices. Students are given 3-minutes to read the passage silently and select a word from the answer choices that restores the meaning of the story. Maze probes have been utilized as reading comprehension assessments for universal screening (Tier 1) and progress monitoring (Tier 2 and Tier 3; Espin, Deno, Maruyama, & Cohen, 1989; D. Fuchs & Fuchs, 1992). The current research study was conducted to further extend the research on the reliability and validity of CBM-maze probes. More specifically, it investigated if there were any differences between 1-minute, 2-minute, and 3-minute time frames, alternate form reliability, concurrent validity, and social validity of the maze probes. Results indicated differences in correct word selections (CWS) between 1-minute, 2-minute, and 3-minute time frames with significant interaction effects noted for the 2-minute maze probe. Alternate form reliability correlation were statistically significant and moderately strong (r = .47 to .71). Concurrent validity correlations between the STAR Reading norm referenced test (computer adaptive reading comprehension test) and CBM-maze probes yielded statistically significant and moderate correlations (r = .30 to .50). Tabulations of the assessment rating scale indicated that students perceived maze probes as acceptable measures for reading comprehension. Implications for practice, cautions in interpreting the results, and future directions are discussed.