Affect Intensity Moderates the Association of Emotional Clarity with Emotion Regulation and Depressive Symptoms in Unselected and Treatment-Seeking Samples

Given the nonclinical nature of this sample, the distribution was checked for normalcy. Using the benchmarks of skewness and kurtosis < 2.0 (Tabachnik and Fidell 2007), the BDI-II scores are considered adequately distributed (skewness = 0.91, SE = 0.22; kurtosis = 0.16, SE = 0.90). We observed three statistically extreme scores at BDI-II values of 22, 23, and 23 (2.59, 2.77, and 2.77 SD above the sample mean, respectively).

Because of the slight skew in BDI-II scores, it was possible the clarity-intensity interaction was driven by just a few participants with elevated depression symptoms. To investigate this possibility, we re-ran our model excluding the three more extreme BDI-II scores. As in the full sample, emotional clarity and negative affect intensity again interacted to predict depression scores, b = − 0.31, SE = 0.11, p = .008, such that the association between emotional clarity and symptoms was statistically significant at low, b = − 1.79, SE = 0.68, p < .01, moderate, b = − 2.71, SE = 0.58, p < .001, high, b = − 3.95, SE = 0.73, p < .001, and very high, b = − 4.87, SE = 0.98, p < .001 levels of negative affect intensity. At very low levels of intensity, emotional clarity was again unassociated with symptoms, b = − 0.86, SE = 0.90, p = .341. To further examine the relevance of this interaction at low levels of depression severity, we repeated analyses in a subsample (n = 101) restricted to the subclinical range (i.e., less than “mild depression,” or BDI-II < 14). Even in this healthier subset, the pattern of results mirrored the full sample. Specifically, emotional clarity was inversely associated with depression severity at moderate, high and very high levels of negative affect intensity (ps < .005; i.e., among the upper 73.27% of this subsample), but unrelated at lower values of negative affect intensity. The differences between these effects were strong enough to yield a marginally significant clarity-by-intensity interaction (b = − 0.18, SE = 0.10, p = .09). Taken together, follow-up analyses show the clarity-intensity interaction was not explained by outliers and appears to operate in both clinical and subclinical participants. This needs to be confirmed with further analyses predicating a broader distribution of depression severity scores (see Study 2).

Some participants skipped occasional items on the emotion regulation questionnaires; we considered data to be missing if the individual had completed fewer than 80% of the items from a given measure. For this reason, and because the ERQ and AAQ-II were added to the questionnaire battery later than the other measures, the sample size differed between models (N = 244 for rumination and non-acceptance models, N = 220 for reappraisal, and N = 168 for experiential avoidance).

Adult participants provided consent for their responses to be used in research; adolescent participants provided assent, and their guardians provided consent.