Great Expectations: Mining State Level Data for Baseline Comparisons to Augment Randomized Experimental Designs

Authors: Michael J. Bryant, Kimberly A. Hammond, Kathleen Bocian, Catherine A. Miller, Michael Rettig, Richard A. Cardullo

We generated a series of baseline (unconstrained) growth models (expected growth in percentage of students scoring proficient or above). These included school and grade level models as well as cohort level models. Our analyses reveal that by holding to the standard AYP criteria, a school or grade level may exhibit a clear separation with gains at the high achievement end potentially obscuring important declines at the lower end of achievement.