Introduces an new approach for measuring the impact of whole school reforms. The approach, based on "short" interrupted time-series analysis, is explained, its statistical procedures are outlined, and how it was used in the evaluation of a major whole-school reform, Accelerated Schools is described (H. Bloom and others, 2001). (SLD)

Despite a growing literature on evaluation methods suitable for applied settings, few investigators address the separation of service and research. This article recognizes obstacles to field studies. The authors specify practical strategies for entering service environments and for designing, implementing, and measuring an experimental…

Human relations services funded under the Emergency School Aid Act were found to have had favorable but generally weak effects on student intergroup relations. Process and impact evaluation should be combined and individual effects should be examined. (Author/CM)

Ongoing program evaluation is examined as an evolutionary process in which the form of evaluation changes in response to the questions to be addressed. As programs evolve over time, so should the evaluation designs. The concept is illustrated with two examples from a large school district in Florida. (Author/CM)

Data analysis approaches for program evaluation under the conditions of natural experiments are explored, using evaluation of the DARE program of drug use prevention among elementary school students as an example. The suggested methodology uses multiple independent analyses to demonstrate the convergence or divergence of program outcomes. (SLD)

The effectiveness of representative teacher evaluation instruments for assessing the eligibility of public school teachers for merit pay was evaluated via a literature review. Elementary school principals' (N=200) ratings of teachers were analyzed. Rating instruments may be useful for making initial decisions about who should be eligible for…

An empirical test of a method of teacher evaluation by predicting student performance and comparing it with actual outcomes is described. Results from 6,249 urban fifth graders in 103 schools suggest that teachers can be evaluated in this way and that the best and worst teachers can be identified. (SLD)