The reliability estimate and standard errors (SEs) of school means are commonly used in determining or assessing distict’s achievements or establishing the “Adequate yearly Progress” for state. Hence, accurate estimations of reliability and SEs are very important. However, in general, the classroom-level variation is ignored when calculating the reliability estimates and SEs of school means. This study was illustrated how the reliability estimates and SEs of school means are influenced by ignoring the class effect. From analysis of variance to demonstrate how an individual student observed score may be decomposed by school effect, class effect, person effect, and residual effect. The results of simulation study and empirical data analysis shown that ignoring classroom-level effects leads an overly optimistic picture of score reliability under classroom effects are nonzero. That is omitting class effects produces series bias in reliability estimates because statistical dependence among scores of students in the same classroom is ignored.
Comments & Questions:
1. By this paper, it can be understood how the influences of the nesting of students within classes within schools. However, this paper pointed that it cannot address the effect of other within-schools organization variables on the reliability of school mean scores. Thus, how to extend this study in general in order to solve other within-schools effects can be investigated in future.