Self-compassion and verbal performance: Evidence for threat-buffering and implicit self-related thoughts

ABSTRACT

Threats to self-esteem can impair well-being directly, e.g., via negative affect, but also indirectly, by impacting performance in valued domains. The present study examined whether self-compassion buffered individuals’ academic task performance from the effects of a self-esteem threat. In addition, this study tested possible effects of self-compassion on implicitly measured self-related thoughts. Participants (N = 333) were randomly assigned to self-esteem threat or neutral conditions, and then either a self-compassion manipulation or an expressive writing (control) condition before completing a set of GRE analogy items. Threat impaired GRE performance in the expressive writing control condition, but not in the self-compassion condition. Moreover, self-compassion appeared to marginally impact implicit non-evaluative self-thoughts, but did not affect evaluative thoughts or implicit self-esteem. The results of this study suggest that self-compassion has benefits for performance and thereby well-being. Future research should further explore the effects of self-compassion on performance and refine understanding of implicit thoughts as possible mechanisms.

KEYWORDS:

• Self-compassion
• self-esteem
• academic performance
• threat

Notes

1. The original target N for data collection was 200, in order to achieve n = 50 per cell. However, the cancellation of another study set to run in the same lab freed up extra time to run additional participants, which we deemed appropriate for an effect previously not demonstrated. No preliminary data analyses were conducted before the study reached its N of 333.

2. The authors did not provide specific wording of the three self-compassion writing prompts they used. As such, an initial wording was generated based on the descriptions provided by Leary et al. (2007), then refined through pilot testing (N = 19) by inspecting pilot participants’ responses and addressing their feedback concerning clarity.

3. English language fluency was included as a control variable to improve the precision of the model, given the diversity of our sample and the fact that English language fluency accounted for 18% of the variance in weighted GRE scores. Omitting this variable led to largely consistent results for the two-way interaction between self-compassion and self-esteem threat, β = .19, p = .050.

4. When not adjusting for language fluency, results were essentially identical: F(1, 303) =  4.72, p = .03, η² = .02.