Predictive Validity Evidence for DIBELS: Correlation to WASL Reading Scores Jack B. Monpas-Huber, Ph.D. Director of Assessment & Student Information
Validity Evidence for DIBELS Primary purpose of universal screeners (like DIBELS, easyCBM) is to predict future performance This prompts a search for evidence of predictive validity, which means correlations with criterion measures (such as state assessments) Previous studies have found strong correlations between DIBELS and state assessments: Following are correlations between DIBELS ORF and WASL scores (Grades 3-6) based on Shoreline data
DIBELS ORF and WASL Reading Standard Grade 3 Nearly all students who meet fall benchmark later met state proficiency standard Students who score below fall benchmark at some risk of not meeting state proficiency standard
DIBELS ORF and WASL Reading Standard Grade 4
DIBELS ORF and WASL Reading Standard Grade 5
DIBELS ORF and WASL Reading Standard Grade 6
What is correlation? What’s a “good” one? Correlation: …is a measure of how consistently two measures sort the same examinees …is a measure of linear relationship between two measures …ranges from zero (no relationship at all) to one (perfect prediction) Typical correlation between WASL scores is in this range A stronger correlation is evidence that two measures are measuring the same thing. We’d expect this of two sets of WASL reading scores. A strong correlation of two different reading measures (WASL and DIBELS) is evidence that both are measuring a lot of the same thing (“reading”) even though each is measuring different reading skills DIBELS-WASL correlations in this range
DIBELS ORF and WASL Reading Standard Correlations Shoreline Public Schools
DIBELS ORF and WASL Reading Standard Grade 3 What is the ORF threshold for meeting WASL standard?
DIBELS ORF and WASL Reading Standard Grade 4
DIBELS ORF and WASL Reading Standard Grade 5
DIBELS ORF and WASL Reading Standard Grade 6
What’s Working?: DIBELS ORF Growth Curves to Study Effectiveness of Reading Programs/Interventions FallWinterSpring Do students in Title schools grow faster? Slower? Does poverty and/or reading instruction predict faster growth (controlling for where kids start)? What does predict rate of learning? Third Grade Cohort (random sample)
What’s Working?: DIBELS ORF Growth Curves to Study Effectiveness of Reading Programs/Interventions What about missing data? It limits our view of kids’ learning. Must statistically “impute” probable growth FallWinterSpring Third Grade Cohort (random sample)
What’s Working?: DIBELS ORF Growth Curves to Study Effectiveness of Reading Programs/Interventions FallWinterSpring Third Grade Cohort (random sample)