Characteristics of participants
The demographic characteristics of the participants are presented in Table 1. The general information form included sociodemographic information (gender, age, education, working years, marital status, position, and number of children). A total of 1627 nurses participated in the study, with a mean age of 31.3 years (SD = 6.0). Among them, 94.7% were female and 5.3% were male. The average number of working years at the institution was 9.3(SD = 6.4). Most participants were married (62.6%) and had a undergraduate degree (89.7%). 76.8% of participants were primary nurses. More than half of the participants had children (56.5%).
-----------------------------------
Insert Table 1 about here
-----------------------------------
Correlations among variables
Table 2 presents the means, standard deviations, and correlations of all the measured variables in our study. First, the results indicated that patient mistreatment was positively correlated with social sharing of negative work events (r=.198, p<.01), work-family conflict (r=.316, p<.01), and emotional exhaustion (r=.361, p<.01) and negatively correlated with perceived organizational support (r=-.319, p<.01). Furthermore, social sharing of negative work events, work-family conflict, and emotional exhaustion were all negatively correlated with perceived organizational support (r=-.193, p < .01; r=-.460, p < .01; and r=-.471, p<.01, respectively).
-----------------------------------
Insert Table 2 about here
-----------------------------------
We used the Harman single-factor test to assess the CMV. Factor analyses showed that the first principal component explained 33.20% of the variance. Before testing our hypotheses, we conducted confirmatory factor analysis (CFA) to confirm the factor structure of our measurement model. As shown in Table 3, the proposed five-factor model fits the data better: c2 = 2492.156, df = 831, Confirmatory Fit Index (CFI) = 0.971, Tucker-Lewis Index (TLI) = 0.970, and root-mean-square error of approximation (RMSEA) = 0.035. Thus, the distinctiveness of our variables is supported [70].
-----------------------------------
Insert Table 3 about here
-----------------------------------
Testing for the mediating effect
Following the mediating effect test procedure recommended by Baron and Kenny (1986) [71], we used SPSS25.0 for hierarchical regression to test our hypothesis. Table 4 presents the results of the hypothesis testing. Hypothesis 1 predicted that there would be a positive association between patient mistreatment and the social sharing of negative work events. As shown in Table 4, the direct path between patients’ mistreatment and social sharing of negative work events was significant, and the estimated coefficient was positive (β=.201, p <.001), thus supporting H1.
In addition, social sharing of negative work events was significantly associated with work-family conflict (β=.206, p <.001) and emotional exhaustion (β=.199, p <.001), supporting the mediating role of emotional exhaustion in the relationship between patient mistreatment and work-family conflict and emotional exhaustion (i.e., Hypothesis 2a). To further test the mediating hypothesis, we calculated the indirect effect of emotional exhaustion and its 95% confidence interval, which was repeated 5000 times using bootstrapping. Bootstrapping is useful for testing indirect effects because it produces a sampling distribution of indirect effects by substituting resampling rather than assuming a normal distribution (Preacher and Hayes, 2008) [53]. The results are presented in Table 5. Social sharing of negative work events significantly mediated the relationship between patient mistreatment and work-family conflict (estimate=.077, 95%CI=[.050,.108]) and emotional exhaustion (estimate=.067, 95%CI=[.043, .094]). Taken together, these results support H2a and H2b.
-----------------------------------
Insert Table 4 about here
-----------------------------------
-----------------------------------
Insert Table 5 about here
-----------------------------------
Testing for the moderated mediation effect
In our conceptual model, perceived organizational support was proposed to moderate the relationship between patient mistreatment, work-family conflict, and emotional exhaustion via social sharing of negative work events. Following Aiken and West (1991), we mean-centered the variables used to form the interaction terms [72]. Table 4 presented the results. The interaction between patient mistreatment and perceived organizational support was significantly correlated with social sharing of negative work events (β=.074, p <.01).
We used the Process plug-in to conduct a simple slope analysis [73, 74]; the results are shown in Table 6. The interaction patterns are shown in Figure 2. The graph shows that when perceived organizational support was low (-1SD), patient mistreatment was positively correlated with social sharing of negative work events (simple slope =.156, p <.001), which was smaller than the coefficient when perceived organizational support was high (+1 SD) (simple slope =.338, p <.001). The moderating effect of perceived organizational support is shown in Figure 2.
-----------------------------------
Insert Table 6 about here
-----------------------------------
-----------------------------------
Insert Figure 2 about here
-----------------------------------
Finally, we used Mplus 8.5 to examine the moderated mediating effects. The results in Table 6 show that the indirect effect of patient mistreatment on work-family conflict via social sharing of negative work events was positive and statistically significant when perceived organizational support was low (estimate =0.033,95%CI=[0.015, 0.055]) and high (estimate=0.070, 95%CI=[0.039, 0.106]). There was a significant difference in indirect effects between high and low perceived organizational support (estimate=0.037, 95%CI=[0.005, 0.074]), supporting H3a. Similarly, the indirect effect of patient mistreatment on emotional exhaustion via social sharing of negative work events was significant when perceived organizational support was low (estimate = 0.029, 95%CI = [0.013, 0.047]) and high (estimate =0.060, 95%CI = [0.035, 0.092]). The difference in indirect effects between high and low perceived organizational support was significant (estimate=0.037, 95%CI=[0.005, 0.074]), supporting H3b.
-----------------------------------
Insert Table 7 about here
-----------------------------------
Additionally, we used the Johnson-Neyman method to depict the continuous confidence intervals of the indirect effects [75]. Figures 3 and 4 show the moderated mediation effect of different POS levels (after standardization), indicating that the effect of patient mistreatment on work-family conflict and emotional exhaustion through social sharing of negative work events was stronger when perceived organizational support was higher.
-----------------------------------
Insert Figure 3 about here
-----------------------------------
-----------------------------------
Insert Figure 4 about here
-----------------------------------