Individuals differ in the way they use rating scales to describe themselves, and these differences are particularly pronounced in children and early adolescents. One promising remedy is to correct (or "anchor") an individual's responses according to the way they use the scale when they rate an anchoring vignette (a set of hypothetical targets differing on the attribute of interest). Studying adolescents' self-reports of their socio-emotional attributes, we compared traditional self-report scores with vignette-corrected scores in terms of reliability (internal consistency), discriminant validity (scale intercorrelations), and criterion validity (predicting achievement test scores in language and math). A large and representative sample of 12th grade Brazilian students (N = 8,582, 62% female, mean age 18.2) were administered a Portugueselanguage self-report inventory assessing social-emotional skills related to the Big Five personality dimensions. Correcting scores according to vignette ratings led to increases in the reliability of scales measuring Conscientiousness and Openness, but discriminant validity and criterion validity increased only when each scale was corrected using its own corresponding vignette set. Moreover, accuracy in rating the vignettes was correlated with language achievement test scores, suggesting that verbal factors play a role in providing both normative vignette ratings of others and self-reports that are reliable and valid.