Results

Tests of Measurement Models

A range of confirmatory factor analyses (CFA) was conducted using Mplus 7.4 to verify how distinctive the variables would be. Four indices were used to evaluate the goodness of fit: the chi-square statistic (χ2), comparative fit index (CFI), Tucker–Lewis index (TLI), and the root mean square error of approximation with associated 90% confidence intervals (RMSEA). CFI ≥ 0.90 and RMSEA ≤ 0.06 indicate a model's acceptable fit to the data. The overall CFA results confirmed that the proposed four-factor model fitted the data excellently (χ2 = 145.72, df = 49, χ2/df = 2.97, CFI = 0.91, TLI = 0.90, RMSEA = 0.08), as shown in Table 2. To examine whether common bias had affected our results, Harman's one-factor model test was performed, inclusive of the 37 items obtained from the same source (i.e., followers) in one model, and a comparison was performed between its model fit indices and the measurement model. The results showed that the one-factor model with a combination of all items was a poor fit for the dataset (χ2 = 917.08, df = 64, χ2/df = 14.33, CFI = 0.42, TLI = 0.25, RMSEA = 0.22). Therefore, we believe that common method variance did not have a significant effect on our data.

Table 2. Results of confirmatory factor analysis.

Model χ2 df χ2 /df CFI TLI RMSEA
4-factor model a 145.72 49 2.97 0.91 0.90 0.08
3-factor model b 235.93 55 4.29 0.78 0.77 0.11
2-factor model c 465.24 58 8.02 0.54 0.51 0.16
1-factor model d 917.08 64 14.33 0.42 0.25 0.22

Note: a charismatic leadership, psychological safety, performance pressure, UPB; b charismatic leadership + psychological safety, performance pressure, UPB; c charismatic leadership + psychological safety + performance pressure, UPB; d charismatic leadership + psychological safety + performance pressure + UPB.


Hypotheses Testing

The analyses were modeled in SPSS 20.0 (SPSS Inc, Chicago, IL, USA) using hierarchical regression. Table 3 presents the means, standard deviations, and correlations among the researched variables. Consistent with the predictions, charismatic leadership is positively related to psychological safety (γ = 0.16, p < 0.01). In addition, psychological safety is positively associated with UPB (γ = 0.20, p < 0.01).

Table 3. Means, Standard Deviations, and Correlations among Variables.

Variable M SD 1 2 3 4 5 6 7
1. Gender a 1.37 0.48 –
2. Age b 1.76 0.81 −0.08 –
3. Education c 2.53 2.22 0.16 * −0.04 –
4. Work tenure d 2.84 1.61 −0.03 0.64 ** 0.09 –
5. Charismatic leadership 2.54 0.68 0.07 −0.01 0.07 0.09 –
6. Psychological safety 4.03 0.45 −0.05 −0.06 0.04 0.01 0.16 * –
7. Performance pressure 3.34 0.56 −0.16 * −0.07 0.03 −0.03 0.12 0.02 –
8. UPB 3.50 0.57 −0.07 0.08 −0.09 −0.02 0.09 0.20 ** 0.05

Note: N = 214. * p < 0.05, ** p < 0.01 (two-tailed); a 1 = male; 2 = female; b 1 = less than 25 year old; 2 = 26–35 year old; 3 = 36–45 year old; 4 = 46–55 year old; c 1 = high school; 2 = junior college; 3 = bachelor; 4 = master; d 1 = less than 1 year; 2 = 1–2 year; 3 = 2–3 year; 4 = 3–4 year; 5 = more than 5 year.

As displayed in Table 4, Model 2, the results indicated that charismatic leadership had a positive relationship with psychological safety (γ = 0.21, p < 0.05). Thus, Hypothesis 1 was supported. As displayed in Model 4, the results indicated that psychological safety had a positive relationship with UPB (γ = 0.23, p < 0.01). Thus, Hypothesis 2 was supported. For Hypothesis 3, the 95% bias-corrected bootstrap confidence intervals were determined for the assumed indirect effect from 5000 bootstrap samples. As shown in Table 5, the correlation between charismatic leadership and UPB via psychological safety was significant (γ = 0.05, 95% CI [0.01, 0.12]), which is a small but significant effect. Thus, Hypothesis 3 was supported.

Table 4. Regression Results.

Independent Variable Psychological Safety UPB
M1 M2 M3 M4 M5 M6
Gender −0.06 −0.07 −0.07 −0.05 −0.04 −0.07
Age −0.08 −0.07 0.12 0.13 0.14 0.13
Education 0.01 0.01 −0.02 −0.02 −0.03 −0.02
Work tenure 0.02 0.01 −0.05 −0.05 −0.05 −0.05
Charismatic leadership 0.21 * 0.18 0.13 0.12 0.10
Psychological safety 0.23 ** 0.23 ** 0.19 *
Performance pressure 0.05 0.05
Psychological safety * performance pressure 0.14 *
R2 0.01 0.04 0.03 0.04 0.07 0.10
∆R2 0.01 0.03 ** 0.03 0.03 * 0.01 0.03 **

Note: N = 214. * p < 0.05, ** p < 0.01 (two-tailed). Standardized coefficients are presented for the linear regression.

Table 5. Indirect Effects of Charismatic Leadership on UPB.

Hypothesis Indirect Effect Mediator Indirect Effect SE 95% Confidence Interval
Hypothesis 2 Psychological safety 0.05 ** 0.03 [0.01, 0.12]

Note: N = 214. ** p < 0.01 (two-tailed).

Hypothesis 4 proposed that performance pressure would moderate the relationship between psychological safety and UPB, such that this positive relationship would be stronger in the presence of high (vs. low) performance pressure. Specifically, performance pressure had a significant moderation effect on the relationship between psychological safety and UPB (γ = 0.14, p < 0.05) in Table 4, Model 6. We plotted this moderation effect in Figure 2. Tests of simple slopes (at +/−1 SD of performance pressure) indicated that when performance pressure was high, psychological safety was significantly positively related to UPB (b = 0.33, t = 4.67, p < 0.01). While the simple slope computed at one standard deviation below the mean, the relationship between psychological safety and UPB was nonsignificant (b = 0.05, t = 0.53, ns). These results suggest that the moderation effect was statistically significant when performance pressure was high. Therefore, Hypotheses 4 was supported.

Figure 2. Moderating Effect of Performance Pressure.

Figure 2. Moderating Effect of Performance Pressure.

We analyzed the contingent indirect effect according to suggestions by Preacher, Rucker, and Hayes. According to Table 6, in the case of higher performance pressure, the indirect effect of charismatic leadership on UPB through psychological safety was of significance (estimate = 0.07, CI [0.01, 0.16]). When performance pressure was low, the indirect effect of charismatic leadership on UPB through psychological safety was insignificant (estimate = −0.01, CI [−0.06, 0.05]). The overall indirect effect of charismatic leadership on UPB through psychological safety was also significant (estimate = 0.06, CI [0.01, 0.15]). Moreover, the indirect effect was stronger at higher levels of performance pressure than at lower levels (differences = 0.08, CI [0.01, 0.17]). Thus, Hypothesis 5 was supported. These effects are illustrated in Figure 3.

Figure 3. Conditional Indirect Effect of Charismatic Leadership and UPB via Performance Pressure.
Figure 3. Conditional Indirect Effect of Charismatic Leadership and UPB via Performance Pressure.

Table 6. Conditional Indirect Effects.

Variables UPB
Estimate SE 95% CI
Low performance pressure −0.01 0.03 [−0.06, 0.05]
High performance pressure 0.07 ** 0.04 [0.01, 0.16]
Index of moderated mediation 0.06 ** 0.04 [0.01, 0.15]
Indirect effect differences 0.08 ** 0.04 [0.01, 0.17]

Note: N = 214. **p < 0.01 (two-tailed).

Note: PP = performance pressure. The indirect effect displays positivity and significance when performance pressure is equal to or greater than 3.50 and equal to or less than 4.14 (on a five-point scale).