Recent developments in the National Assessment of Educational Progress (NAEP) have addressed criticisms of previous assessments and have focused on state education departments as the major clients for test results. NAEP currently indicates progress in academic learning but does not define which aspects of schools are responsible for successes and failures. Recent reports indicate that NAEP plans to assess the association of input variables and mediating variables and school performance, in order to make the program more useful to policymakers. Sophisticated methods such as path analysis and structural modeling will permit many variables to be analyzed simultaneously. The variables to be studied should be actionable or practical to carry out, should reflect a reasonable theory about what matters in producing achievement, and should specify a time scale of expected effects. Actionable variables include teacher educational background and salary, instructional time, textbooks, class size, classroom homogeneity, homework, tracking, and courses taken. Mediating and interacting variables include attitudes, social background, and academic ability. Additional attention to data analysis is suggested as the most important of the recommendations. The appendix summarizes criticisms of NAEP and its current plans. (GDC)