The Reliability of Self-Reported GPA in Educational Research: A Comparison of Self-Reported and Officially Recorded Data

Researchers use often self-reported grades as indicators of students' educational achievement because of difficulties in obtaining the official records. However, the results of such studies are questionable due to doubts regarding the reliability of self-reported grade data and factors that may lead to distortion. In this study, we help to resolve this issue by comparing self-reported grade point averages (GPAs) of 926 eighth grade Romanian students from the previous academic year with their official GPAs. At the descriptive and practical levels, we conclude that the distortions are positive-overreporting is more frequent than underreporting-and that their limited magnitude allows for the use of self-reported grades for the usually description i.e. of achievements in a population of students but not for individual students' diagnostics. At the explanatory level, the distortions are non-random, and there are two main causal mechanisms. First, overreporting is a cognitive and commitment-based error. Second, overreporting is an effect of social desirability, according to which students are expected to perform well based on their social characteristics. However, this second causal channel is not visible in bivariate tests because it is statistically suppressed by the main mechanism and the correlation of actual GPA with gender and socioeconomic status (SES - parents' education and class) variables.