ABSTRACT
Expressing (vs. withholding) forgiveness is often promoted as a beneficial response for victims. In the present research, we argue that withholding (vs. expressing) forgiveness can also be beneficial to victims by stimulating subsequent transgressor compliance – a response that is valuable in restoring the victim’s needs for control. Based on deterrence theory, we argue that a victim’s withheld (vs. expressed) forgiveness promotes transgressor compliance when the victim has low power, relative to the transgressor. This is because withheld (vs. expressed) forgiveness from a low-power victim elicits transgressor fear. On the other hand, because people are fearful of high-power actors, high-power victims can expect high levels of compliance from a transgressor, regardless of whether they express forgiveness or not. A critical incidents survey (Study 1) and an autobiographic recall study (Study 2) among employees, as well as a laboratory experiment among business students (Study 3), support these predictions. These studies are among the first to reveal that withholding forgiveness can be beneficial for low-power victims in a hierarchical context – ironically, a context in which offering forgiveness is often expected.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. Without including these six respondents, the results of the focal two-way interaction remained similar to those presented in the main text (F(2, 87) = 2.77, p = .07, η2 = .06).
2. Previous studies have shown that perceived sincerity of the other party’s gestures and moral entitlement of two parties are important mediators in the reconciliation process (Zheng et al., Citation2016; Zitek, Jordan, Monin, & Leach, Citation2010). To rule out that these mediators explain our findings, we measured sincerity perceptions of victim responses and transgressors’ feeling of moral entitlement (Zheng et al., Citation2016; Zitek et al., Citation2010) and tested for their potential mediating role using Hayes’ PROCESS macro (Model 8, 5,000 bootstrap resamples) by including sincerity, entitlement, and fear as mediators. Results revealed that only fear was a significant mediator.
3. Given that whether or not the transgressor apologized may shape the extent to which they fear punishment and retaliation as well as their subsequent compliance, we measured whether the transgressor has apologized (1 = yes, 0 = no). Correlation results indicated that apology was not related to fear (r = −.02, p = .84) or compliance (r = −.07, p = .46).
4. Without controlling for gender, a victim power (high vs. low) × victim response (expressing forgiveness vs. withholding forgiveness) ANOVA on transgressor compliance revealed that the main effect of victim power was significant (F(1, 85) = 12.36, p < .01, η2 = .13). The main effect of victim response was not significant (F(1, 85) = .33, p = .57, η2 = .00). Furthermore, the two-way interaction was not significant (F(1, 85) = 3.77, p = .055, η2 = .04). The simple effects of victim responses were not significant for low-power victims (F(1, 85) = 3.00, p = .09, η2 = .03) and for high-power victims (F(1, 85) = .99, p = .32, η2 = .01). The simple effects of victim power were significant for victims who expressed forgiveness (F(1, 85) = 13.53, p < .01, η2 = .14) but was not significant for victims who withheld forgiveness (F(1, 85) = 1.38, p = .24, η2 = .02).
5. Without controlling for gender, a victim power (high vs. low) × victim response (expressing forgiveness vs. withholding forgiveness) ANOVA on transgressors’ fear revealed that the main effects of victim response (F(1, 85) = .76, p = .39, η2 = .01) and victim power (F(1, 85) = 2.74, p = .10, η2 = .03) were not significant. Importantly, there was a significant interaction of victim power and victim response on transgressor fear (F(1, 85) = 9.62, p < .01, η2 = .10). The simple effects of victim responses were significant for low-power victims (F(1, 85) = 7.49, p < .01, η2 = .08) but not significant for high-power victims (F(1, 85) = 2.63, p = .11, η2 = .03). The simple effects of victim power were significant for victims who expressed forgiveness (F(1, 85) = 10.28, p < .01, η2 = .11) but not significant for victims who withheld forgiveness (F(1, 85) = 1.16, p = .28, η2 = .01).
6. Scholars increasingly recognize that to evaluate an effect, one should not consider single studies in isolation but instead evaluate the totality of the evidence across studies (e.g., Lakens & Etz, Citation2014; Tuk, Zhang, & Sweldens, Citation2015). To do so, scholars recommend within-paper meta-analyses (e.g., Cumming, Citation2014). We therefore conducted an internal meta-analysis on the three studies in this paper. Specifically, we used van Rhee, Suurmond, and Hak (Citation2015)’s software to compute the meta-analytic effect size and its 95% confidence interval for the victim power × victim response interaction on transgressor compliance (number of studies k = 3, number of participants n = 294). This analysis revealed significant support for our prediction that victim power and victim response interact to influence transgressor compliance (d = .26, 95% CI:[.13, .40], Q(3) = .68, p = .71).