Correlations between cortisol level and internalizing disposition of young children are increased by selecting optimal sampling times and aggregating data.

To determine the optimal time for testing the association between personality traits and cortisol levels, and the number of sampling days needed for reliable results to be obtained for preschool-aged children, ambulatory salivary sampling for cortisol was performed on 39 preschoolers at four different times for 10 consecutive school days. A repeated-measure ANOVA revealed a significant day effect, time effect, and day by time interaction. Further tests of day effect revealed that within-subject cortisol levels differed significantly in the early morning, early afternoon, and late afternoon cortisol data, but not in the late morning; group average levels of cortisol assessed in the early mornings are significantly different among the days, so are the ones measured in early and late afternoon. Mid-afternoon cortisol levels showed the most significant day effect and the highest correlation with internalizing disposition. In addition, the probability of finding significant correlations between cortisol levels and internalizing disposition increased with a greater number of aggregated cortisol measures. The tentative conclusion is that to avoid under-representation of the correlations between cortisol levels and personality variables, both sampling time selection and data aggregation are important.