Programevaluation

Program Evaluation: School Performance by Instructional Spending Levels

-prepared by Dr. James E. Storbeck

The notion of a program evaluation in the context of performance assessment is important, in that it allows one to examine the levels of performance associated with different programs (or, policies). Furthermore, using nonparametric tests, one can make statements as to the statistical significance of emergent differences in performance.

In order to illustrate this capability, we turn to the Performance model results from the [State] Department of Education School Performance Project. The particular proposition being tested in this analysis relates to the often-cited concern that there are not enough spending dollars "reaching the classroom." The question to be addressed in our context is this: as classroom-oriented (i.e., instructional) dollars increase as a percent of a school's budget, does performance improve?

In this example, then, efficiency values are sorted by instructional spending shares of school budget, and assigned to one of three equally-sized groups. Each group of schools, then, represents a different "program" to be evaluated. The ordering of the groups is from "Low" (instructional dollars as percent of total budget) to "High." If school performance (as measured by efficiency value) does not improve with the level of instructional dollars share of budget, the mean performance levels for all groups should be equal. Using a program evaluation format, accompanied by simple nonparametric statistics, we can indeed test this (null) hypothesis of equal means.

Results

Table 1 reports the average efficiency value by instructional spending shares. Let us first focus on the three means in the "overall" efficiency column. These means, which are simple averages of school efficiencies, would be equal for all three groups if the instructional share of budget did not "make a difference." Clearly, such is not the case, as mean...