The descriptive statistics used means ± standard deviation for quantitative variables and percentages for qualitative ones. Comparisons between groups of students (responders vs. non-responders) and short vs. long dreams were performed using chi-square tests for qualitative variables and Student’s t-tests for quantitative variables. The determinants of the individual mean grade (result of the competition) were assessed using two steps: a univariate analysis was performed using Spear- man’s correlations for continuous or ordinal variables, and Student’s t-tests or ANOVAs were performed for qualitative vari- ables with two or more than two values, respectively; and a multivariate linear regression using a backward variable selection method, which initially included all variables with a p-value lower than 0.10 in the univariate analysis and ulti- mately retained all variables with a p-value lower than .05. We chose to perform the selection method in multivariate anal- yses because the multivariate p-value (the final one) is impossible to predict from the univariate one (the p-value of the first step). All tests were two-sided, and computations were performed using SAS version 9 (SAS Institute Inc., Cary, NC, USA).