ABSTRACT A common method of studying cross-situational variation in personality involves asking people to describe their behavior in several different contexts. Although this approach is intuitively appealing, it introduces a great deal of redundancy into the measurement process and may affect the interpretation of contextualized self-reports. Specifically, when participants are asked the same questions repeatedly in a single questionnaire, they may be compelled to focus on how their behavior is different. We tested this hypothesis by experimentally manipulating the number of contexts that were included in a measure of role-related personality. In 2 experiments (Study 1 N=377, Study 2 N=524), we found that multiple-role questionnaires produced greater variation in trait levels across roles, larger differences between general and role-specific ratings, and weaker correlations between general and role-specific ratings than single-role questionnaires. These findings illustrate how the measurement process can have an effect on the variability of responses to contextualized self-reports.