This article looks at the methodological problems created by incorrectly using between-class data when looking at effects generated by individual students. The grade/evaluation association found in the student evaluation of teaching was used as a case example. A theoretical framework to understand the contributions to the relationship by both between-class data (leniency) and within-class data (reciprocity) was created that shows how the two are confounded when looking only at between-class effects. The findings suggest that within-class data is the appropriate starting point when investigating associational problems with either within-or between-case data. (Contains 2 tables and 2 figures.)