School A: Highest IS School D: Low IS, Better than exp gain School C: Low IS, High Gain School B: Highest Gain School E: Med IS, Lower than exp gain Criteria You Choose Will Yield Different Conclusions About School Performance
Example of CRESST Model: Latent Variable Hierarchical Model Level 1: Two Time-point Series Obtaining initial status and gain for each student Level 2: Student Level Gain for student is modeled as function of his/her initial status (& SES) Level 3: School Level Gain for school is modeled as function of school’s initial status (& SES)
CRESST Study 1: Comparing AYP and VA Classifications Examined gains for 3 Performance Subgroups within each school Subgroups defined by initial starting point Hi Performers: 2 SDs above mean Average: Mean initial starting point Low Performers: 2 SDs below mean Also possible to define performance subgroups based on a set of absolute values (e.g., 10 pts above and below mean)
3 AYP Schools: Distribution of Student Gain
2 AYP Schools: Distribution of Student Gain
2 Non-AYP Schools: Distribution of Student Gain
CRESST Study 2: Adjusting for Background Characteristics in VAM School effectiveness literature still searching for appropriate ways to adjust for background characteristics Type A: adjustment of student background (S), yet no adjustment of school-level contextual effects (C) and school policies and practices effects (P) Type B: adjustment of student background (S) and contextual effects (C)
Correlations between Model Results No adj Stud IS Stud SES Stud IS_SES Stud IS_SES & Mean IS Stud IS_SES & Mean IS_SES 1 0.95 0.99 0.98 0.93 1.00 0.97 Stud IS_SES & MeanIS 0.90 0.79 0.87 0.92