I have also shown that self-control predicts report card grades (and changes in report card grades over time) better than does measured intelligence (Duckworth & Seligman, 2005). To strengthen the inference that self-control causes, and does not merely predict, report card grades, I repeatedly sampled self-control and report card grades in a separate sample of students over a four-year period and then developed an innovative application of hierarchical linear modeling (HLM) to analyze the data (Duckworth, Tsukayama, & May, 2010). Specifically, I used self-control as a time-varying level-1 covariate to predict GPA six months later, and then reversed the same model by using GPA as a time-varying level-1 covariate to predict self-control six months later. Within-individual changes in self-control over time predicted subsequent changes in GPA but not vice-versa. The evidence supporting a causal role for self-control was not moderated by level-2 covariates including IQ, gender, ethnicity, or income. Further analyses ruled out the potential time-varying confound of self-esteem.