Impact of the Understanding Science Professional Development Model on Science Achievement of English Language Learner Students
Analysis plan
To estimate program impacts, the outcomes for students and teachers in treatment classrooms are compared with those for their counterparts in control classrooms. In addition, multilevel regression models, which account for school and classroom clustering, are used to analyze Understanding Science's effect (Goldstein 1987; Raudenbush and Bryk 2002; Murray 1998). Each impact analysis will control for baseline measures (pretests) of outcome variables and other student- and teacher-level covariates. These models will be used to estimate impacts on student performance for all participating students and for English language learner students.
In addition to examining the main effects of the program on student performance, exploratory analyses will be used to examine program impacts for certain student subgroups: English language learner and English-proficient students, low- and high-performing students, students in different racial/ethnic groups, and students from low-income households.
Procedures described by Schochet (2008) are used to account for testing multiple hypotheses involving primary outcome variables. Five outcome domains were delineated for this study: student content knowledge, student science achievement, teacher content knowledge, teacher instructional strategies, and teacher confidence in ability to teach force and motion. Multiple comparison procedures are used for each outcome domain to reduce the probability of finding statistically significant program impacts when they are due to chance alone.