Grant

Procedures that Optimize the Reliability and Validity of Classroom Observations

Does the length of time/frequency of classroom observations impact the reliability and validity of the CLASS measurement system? How stable are variance estimates in a G-Study, and do they produce replicable results in a D-Study?

Investigators will randomly assign classroom observations to different coding procedures, compare reliabilities across coding procedures and CLASS domains, and examine whether coding procedures affect the validity of inferences about the CLASS. They will also conduct a Monte Carlo simulation to inform sample size requirements for G-Studies. They will cross-validate D-Study results by comparing conjectured reliability estimates to actual estimates.