ABSTRACT
When apologizing to victims of transgressions, people may assign the agency for harm to themselves (“I’m sorry I offended you”), to the act (“I’m sorry it offended you”), or omit agency altogether (“I’m sorry you were offended”). They also may acknowledge or question the victim’s harm by the choice of conjunction used to introduce an explanatory subordinate clause (“I’m sorry that you were offended” vs. “I’m sorry if you were offended”). Participants (n = 1,118) were randomly assigned to read 1 of 12 versions of a scenario wherein someone commits a transgression (e.g., breaks a promise) and then apologizes. Versions varied on degree of offense (minor or major), apology agency assignment (transgressor, transgression, or omitted), and conjunction choice (“that” or “if”). Results indicated that apologies assigning agency to the transgressor or transgression invited more favorable perceptions of the transgressor than those omitting agency. However, there was no reliable main effect nor interaction involving subordinating conjunction choice. Our findings demonstrate the impact of strategic word choice in apology on victims’ perceptions of transgressors.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. Investigation of normality of perceived harm (Kurtosis = 4.37 for minor transgression and 9.09 for major transgression) and perceived intention (Kurtosis = 4.34 for minor transgression) in Scenario 2 revealed the distributions to be substantially skewed. Given the nature of the offense and the robustness of the ANOVA against violation of the normality assumption (Schmider, Ziegler, Danay, Beyer, & Bühner, Citation2010), no data transformation was performed. Results of Levene’s tests in Scenario 1 indicated that the assumption of homogeneity of variance was met for all dependent variables. In Scenario 2 the assumption of homogeneity of variance was unmet for perceived harm [F(11, 432) = 2.04, p = .023], perceived intention [F(11, 432) = 6.69, p < .001], perceived remorse [F(11, 432) = 7.19, p < .001], and likelihood of punishment [F(11, 432) = 2.70, p = .002]. In Scenario 3, Levene’s test indicated equal variance for all dependent variables but perceived responsibility taken [F(11, 312) = 2.07, p = .022]. Because ANOVA is robust against violation of the equal variance assumption when group size is approximately equal (i.e., the largest group size does not exceed 1.5 times of the smallest group size) and that unmodified Levene’s test is neither the best nor the worst in terms of power and robustness compared with other homogeneity of variance tests such as modified Levene and Z-variance (Lee, Katz, & Restori, Citation2010), we made the decision of accepting that the assumption was unmet and proceeded with the univariate ANOVA cautiously.