ABSTRACT
What are the social signals of emotional tears? This question has fascinated scholars ever since Darwin. Studies have suggested several interpersonal effects of emotional tears. A recent article presented evidence in three studies that tearful individuals are not only perceived as warmer but also less competent than their non-tearful counterparts. However, the competence effect was relatively small, and a recent replication failed to find such an effect in two different populations while the warmth effect was replicated. This questions the generalizability of the effect of tears on perceived competence. To test whether individuals expressing emotional tears are really perceived as less competent and what boundary conditions such an effect might have, we specified a decision tree of three different studies in which we test differences between the original reference study and the replication. We replicated previous findings of the perceived (in)competence in Study 1 (n = 531) but observed a considerably smaller nonsignificant effect when proceeding to Study 2 (n = 471), which increased the number of stimuli. The earlier and now repeated replication failure can therefore likely be attributed to an increased variation in stimulus material. We conclude that there is not enough evidence to argue that one social outcome of tears signals a relative lack of competence as the effect seems to depend on the specific stimuli used.
Acknowledgment
We are grateful to Marco Anelli for allowing us to use his photographs.
Disclosure statement
No potential conflict of interest was reported by the authors.
Supplementary material
Supplementary material can be accessed here.
Author contributions
All the authors conceived and designed the studies. JZ wrote the first draft, and all authors revised it.
Notes
1. Note that in this research, we follow the prior work on crying, testing the perception of tears shed out of sadness or being moved. Copying the original studies, we do not add information about the context of tears. Thus, as we study tears that could be perceived as negative, but also positive, we use the term emotional tears. Although people might sometimes also shed tears when laughing (Niedenthal, Mermillod, Maringer, & Hess, Citation2010), this is not what we focus on in this research.
2. Reanalyzing Zickfeld and Schubert’s (Citation2018) data only using the two faces employed by Van de Ven et al. (Citation2017) still indicates no significant effect for perceived competence. However, the power of this approach might be too low to detect such an effect (number of cases: United States, n = 188; Norwegian, n = 94).
3. Reanalyzing ZS’ data only using the 10 faces employed in Study 1 by VMV still indicates no significant effect for perceived competence in the United States (n = 349) or Norwegian sample (n = 211). However, the power of this analysis might be too low to detect such an effect.
4. Note that although previous findings have pointed at the possibility of an influence of target gender, findings have been inconclusive and generally suggest no difference (see ). As the theoretical basis does not explicitly predict specific gender differences, we therefore rely on the overall effect as our estimate throughout all studies.
5. Please note that we report eta squared and not partial eta squared throughout the manuscript.
6. Nevertheless, we explored the competence effect in the set of 10 targets used by VMV. Again, we observed no significant main and interaction effects. The tearful targets were rated as slightly less competent, B = −.14, t(459) = −1.59, p = .11, and no significant effect of target gender, B = −.19, t(8) = −.83, p = .43. However, we exploratorily repeated the model for each target gender separately. Again, we observed a small effect that tearful male targets were rated as less competent (Mno tear = 3.87, SDno tear = 1.12 vs. Mtear = 3.56, SDtear = 1.23; d = −.26 [−.40, −.12]), while this effect was weaker for female targets (Mno tear = 3.52, SDno tear = 1.20 vs. Mtear = 3.42, SDtear = 1.18; d = −.08 [−.24, .08]).
7. Performing equivalence tests, the effect size for female targets was statistically different from our smallest effect size of interest of d = ±.30 (TOST, 90% CI [−.27, −.07]), while the effect for male targets did not differ (TOST, 90% CI [−.37, −.10]).