Emotion regulation choice in an evaluative context: the moderating role of self-esteem

ABSTRACT

Evaluative contexts can be stressful, but relatively little is known about how different individuals who vary in responses to self-evaluation make emotion regulatory choices to cope in these situations. To address this gap, participants who vary in self-esteem gave an impromptu speech, rated how they perceived they had performed on multiple evaluative dimensions, and subsequently chose between disengaging attention from emotional processing (distraction) and engaging with emotional processing via changing its meaning (reappraisal), while waiting to receive feedback regarding these evaluative dimensions. According to our framework, distraction can offer stronger short-term relief than reappraisal, but, distraction is costly in the long run relative to reappraisal because it does not allow learning from evaluative feedback. We predicted and found that participants with lower (but not higher) self-esteem react defensively to threat of failure by seeking short-term relief via distraction over the long-term benefit of reappraisal, as perceived failure increases. Implications for the understanding of emotion regulation and self-esteem are discussed.

KEYWORDS:

• Emotion regulation
• choice
• distraction
• reappraisal
• self-esteem

Acknowledgements

We wish to thank James J. Gross for his help and support, and Prof. Daniel Yekutieli and Noa Haas for statistical consultation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Because this exclusion criterion may be somewhat subjective in nature, we also re-ran the main analyses including these five participants, and results remained essentially unchanged (i.e. the interaction between self-esteem and performance, OR = 0.97, 95% CI = (0.95, 0.99), p < .001, as well as the follow-up analyses for lower self-esteem individuals, OR = 1.2, 95% CI = (1.03, 1.41), p < .02, and higher self-esteem individuals, OR = 0.89, 95% CI = (0.77, 1.04), p = .14).

2. To provide further support for the adequacy of our final sample size, we carried out a simulation-based power analysis for our generalised linear mixed model (see Johnson, Barry, Ferguson, & Müller, 2015 for a full description of the method). Specifically, 1000 simulated data sets were generated randomly, applying the observed effect size of the interaction between self-esteem and performance (beta = −.032) and the present sample size. Mean and standard deviation of self-esteem and performance were taken from the observed data. This analysis indicated that the power of the observed results was high, manifested in an 84% chance of detecting the central interaction described in our article with the present sample size and an alpha of .05.

3. The dimensions (affability, alertness, articulateness, capability, competence, confidence, considerateness, creativity, honesty, innovation, insightfulness, intelligence, knowledge, peacefulness, persuasiveness, posture, responsibility, sensitivity, tolerance, and trustworthiness) were derived from previous research on evaluative feedback (Kernis & Johnson, 1990; Mansell & Clark, 1999), as well as general parlance.

4. Results are reported without controlling for pre-speech anxiety ratings; however, the main results remain essentially unchanged when including pre-speech anxiety as a covariate (i.e. the interaction between self-esteem and performance, OR = 0.97, 95% CI = (0.95, 0.99), p < .002, as well as the follow-up analyses for lower self-esteem individuals, OR = 1.33, 95% CI = (1.11, 1.63), p < .005, and higher self-esteem individuals, OR = 0.88, 95% CI = (0.75, 1.02), p = .09).

5. At the trial level (within persons), perceived performance was negatively correlated with single-trial post-regulation anxiety ratings (r = −.29, p < .05), suggesting that reduced perceived performance (failure) is associated with increased negative feelings. Single-trial post-regulation anxiety ratings were included to heighten the salience of our instruction to base regulatory choices on their impact on emotional responses (see Scheibe, Sheppes, & Staudinger, 2015 for a similar inclusion). Post-regulation ratings are not discussed further, because in choice studies they are un-interpretable with regard to differential effectiveness of employing distraction and reappraisal. Specifically, because participants freely choose between reappraisal and distraction, it is impossible to control for initial potential intensity differences between distraction and reappraisal chosen stimuli (see also Scheibe, Sheppes, & Staudinger 2015 for a thorough discussion).

6. Given that we wished to explore the future possibility of repeating the present design with relevant clinical populations suffering from social anxiety disorder, in the end of the experiment participants also completed the Brief Fear of Negative Evaluation (BFNE; Leary, 1983), which measures fear of negative evaluation symptoms. Since these social anxiety symptoms are directly related to self-evaluation (as also manifested in the high correlation with self-esteem, r = −.60, p < .001), when BFNE was entered as a moderator to our multilevel logistic regression analysis (instead of self-esteem), it significantly interacted with performance (OR = 1.01, 95% CI = (1.005, 1.02), p < .002). Similar to the pattern of results observed with the self-esteem measurement, follow-up analyses showed that participants with higher social anxiety symptoms were more likely to distract as perceived poor performance (failure) increased (OR = 1.25, 95% CI = (1.09, 1.44), p < .002), whereas those with lower social anxiety symptoms appeared to be less affected by their perceived performance (OR = 0.9, 95% CI = (0.77, 1.05), p = .18). In addition, we also administered the State Anxiety Questionnaire (STAI-S; Spielberger, 1983), to be able to control for its potential influence on self-esteem). Importantly, the interaction between self-esteem and performance remained significant when controlling for state anxiety levels (OR = 0.97, 95% CI = (0.95, 0.99), p < .002). Participants also completed the emotion regulation questionnaire (Gross & John, 2003).

7. When defining the effects of self-esteem and performance as random, the interaction between self-esteem and performance remained essentially unchanged (OR = 0.96, 95% CI = (0.94, 0.99), p < .002).

8. ICCs reflect the proportion of variability in a repeated measure due to between-person variation. For binary outcomes, ICCs are computed as the proportion of between-person variance over the total variance, where the trial-level variance is estimated as π²/3 – the standard logistic distribution’s variance.

9. Reverse-coding was done to facilitate interpretation of the ORs (to be greater than 1.0) for focal effects.

Additional information

Funding

Gal Sheppes is supported by Israel Science Foundation [grant number 1130/16].