Because public schools do not randomly assign students and teachers across schools (a methodological utopia), multilevel evaluation methods which account for student and school contextual and practice variables in their natural settings provide the most rigorous means for showing empirically what is actually happening in school classrooms. However, no statistical methodology can make up faulty design or bad data. This paper presents some important practical issues regarding data handling for multilevel analysis methodology. Also presented are important modeling design issues that need to be considered when applying hierarchical linear models (HLM) to the measurement of schools and for determining which factors impact the value schools added to students' achievement. Data handling issues that must be considered in HLM are determining the unit of analysis, variable selection and measurement standards, and harvesting raw data from school district sources. The second section of the paper discusses the HLM model and the importance of centering, the estimation of school effects, and the empirical Bayes estimation procedure. (Contains 1 table, 1 figure, and 14 references.) (Author/SLD)