Some universities administer standardized tests to estimate school effects on student learning. To that end, many of those institutions choose to gather cross-sectional data (freshmen and seniors in the same academic year) rather than longitudinal data. The effects of this decision on institutional assessment results are generally unknown. The study reported here compared longitudinal and cross-sectional school effect estimates using data from 25 schools that administered the Collegiate Learning Assessment (CLA), an open-ended test of critical-thinking skills. Four different statistical methods were used to estimate school effects, including value-added models commonly used in K-12 and higher education. Correlations between longitudinal and cross-sectional results averaged 0.47, revealing that data-collection methodology had a large impact on the relative standings of school effect estimates. The effect of using different statistical models was much smaller. Effect sizes, reflecting longitudinal and cross-sectional differences between freshmen and seniors, were similar on average, but differed greatly at a few schools. Effect sizes varied across schools, suggesting that some schools are more successful than others at increasing students' critical-thinking skills. [Paper presented at the 2015 National Council on Measurement in Education Annual Meeting, Chicago, IL.]