Aggregated Randomly Generated Math Curriculum-Based Measurements for Middle School Students: Reliability, Predictive Validity, and Cut Score Precision.

Approximately, 283 sixth-grade students completed four randomly generated Math Curriculum–Based Measures (M-CBMs) from interventioncentral.org twice during a fall semester. These M-CBMs measured foundational skills in addition, subtraction, multiplication, and division. We found that using mean fluency or accuracy scores from four M-CBMs produced more reliable test–retest estimates (e.g., r = .86–.88) in comparison with using individual scores (e.g., r = .55–.73). We also found that aggregated (mean or median) and individual M-CBM scores were strongly correlated (e.g., r = .82–.85) with the Palmetto Achievement Challenge Test (PACT) math section, an annual state standardized test. However, only aggregated M-CBMs produced efficient cut scores useful in detecting students who would pass or fail the PACT.