Abstract
Individuals sometimes are hurt by the comments of others, and residual feelings of hurt are experienced after the episode. Because of individuals’ tendency to ruminate, we argue that residual hurt and its consequences may be especially common among individuals who tend to take conflict personally (TCP). We hypothesized that TCP would be positively related to motivations to avoid and to seek revenge against offenders, and these relationships will be mediated by the link between rumination and residual hurt. To test our hypotheses, we conducted a survey among undergraduates about how they reacted to a hurtful message. We confirmed our hypotheses.
Notes
Note. N = 130.
a Reliability coefficients for multi-item scales reported along diagonal.
b Correlations between continuous measures are Pearson correlations and those between dummy coded and continuous measures are point-biserial correlations.
c Time Elapsed was measured in months.
d 0 = male; 1 = female.
e 0 = current relationship; 1 = terminated relationship.
f 1 = friend; 0 = romantic partner; 0 = family; 0 = other.
g 1 = family; 0 = friend; 0 = romantic partner; 0 = other.
h 1 = romantic partner; 0 = friend; 0 = family; 0 = other.
*p < 0.05. **p < 0.01.
Note: N = 130.
Number of bootstrap samples for bias corrected bootstrap confidence intervals = 5000.
a = Unstandardized coefficients.
b = 95% Confidence intervals.
c = 95% Bootstrapped confidence intervals.
*p < 0.05. **p < 0.01. ***p < 0.001.
Note: N = 130.
Number of bootstrap samples for bias corrected bootstrap confidence intervals = 5000.
a = Unstandardized coefficients.
b = 95% Confidence intervals.
c = 95% Bootstrapped confidence intervals.
*p < 0.05. **p < 0.01.
Mediation also can be tested through structural equation modeling (SEM). We chose to use PROCESS for two reasons. First, some scholars argue that to appropriately use SEM, a relatively large ratio of subjects to parameters should exist. Kline (Citation2011) noted that a ratio of at least 20 subjects per every one parameter should exist and Schumacker and Lomax (Citation2010) noted that for some models, the ratio may be as high as 500 to 1. Our sample size falls short of those standards. Second, although SEM provides a means of testing all possible relationships among variables in a model, not all of the relationships are hypothesized and they could influence model fit. PROCESS provides a test of the predicted mediated relationships without including other unhypothesized ones. We acknowledge that by not using SEM, we sacrifice the ability to adjust for measurement error, which may attenuate our effect sizes.