Reference no: EM133129283

We learn there are a few reasons to be concerned about interpreting estimates longitudinally as causal. The reasons are the following:

-In Grades 1-3 even regular classes without aides had a part time aide.

-Students in regular classes and regular with aide classes were randomly re-assigned between these two types of classes after kindergarten while those in small classes remained in the same small class.

-New students entered in first grade who did not attend kindergarten, people repeated grades, families moved. New students randomly assigned but limited slots for "small size" classrooms. "unbalanced allocation"

-Concern about the fact that randomly assigned class size is not always the actual class size they end up in. But they find that only 0.3% of. Students were not enrolled in class type they were randomly assigned to.

-Don't have the baseline test scores before the groups were "treated" by class size. i.e. we don't know what the original test scores were for all the students in the experiment so we can't compare if the ones who were randomly assigned to small classes had similar test scores to those in other groups.Describe how two of these reasons could bias the results.