This paper highlights measurement issues faced when attempting to assess and interpret results of a school improvement project. Based on the assumption that to measure effectiveness, one must measure a wide variety of school factors, the paper presents a broad perspective on measurement problems and dilemmas in analyzing norm-referenced test data and data obtained through interview, self appraisal, and observation concerning ll7 elementary schools. Trend analysis, two forms of residual gains analyses, traditional ranking, and expert judgment methods are compared. Data suggest that school level residual analysis appears to provide the best approach to selecting schools. The individual level residual scores yield a list which overlaps with the school level approach. Trend analysis is the most conservative and yields the fewest schools (which are also identified by residual score analyses). Expert opinion does not correlate positively with residual or trend analyses. Analyses indicated few consistencies over time. The authors conclude with two alternatives--either schools are not consistent in their impacts from year to year or their metric is suspect. (Author/CM)