Sample Profile
Data of the present study were collected from 435 participants. Descriptive statistics showed that the demographic variables included (1) gender, (2) age, (3) education level, (4) hierarchical level (5)tenure and (6) organizational scale in this study. Among the respondents (N = 435), almost 57% were female employees and 43% were male. In terms of age, the majority were in their 20s (55%) and 30s (32%). As for the educational level, 73% of the respondents had bachelor’s degree and 23% from graduate school. In terms of hierarchical level, 83% of the respondents were in general staffs, 12.4% were in lower supervisory positions and 4.6 % were higher than middle managerial level. The tenure of employees was distributed across the following categories: below 5 years (56%), 5-10 years ( 21%), 10-20 years (18%), 20-30 years (4%) ,and over 30 years (1%). The organizational scale also divided into different sizes, 67% were worked in middle-size organizations (below 500 employees), 15.2 % worked in large organizations (500–3,000 employees), and 17.9% worked in very large organizations (more than 3,000 employees).
Table 4. 1.
Descriptive Statistics on Sample Characteristics (N = 435)
Variable Categories Count Percentage (%)
43
Tenure in current job position Below 5 5-10
Company Scale Below 300
300-500
44
Pearson’s Correlation Analysis
Correlation analysis is conducted in order to examine and investigate the direct correlations (or relationship) between variables, and it is used to measure the strength of relationships between two continuous variables (Gray & Kinnear, 2012). By doing so, the effects of independent variables on dependent variables can be tested, verifying whether they are positively or negatively correlated with each other. The correlation analysis of this study is performed by IBM SPSS 23.0. When correlation values are above 0.75 (Vernon &
Mior, 1991), it may imply the existence of multicollinearity between the tested variables.
When the correlation is high, it represents that there is a strong relationship between two variables. The figure of correlation should between 1 and -1.
The result of correlation provides initial test of the relationships among different variables. The result demonstrated that Work Engagement was negatively correlated with Acquiescent Silence (r = -.23, p < .001), Defensive Silence (r = -.19, p < .01), and Opportunistic Silence (r = -.15, p < .01). On the other hand, work engagement was positively correlated with Mentoring Relationship (r = .65, p < .01) and Employee Well-Being(r = .77, p < .01). These results showed preliminary directions and correlations are consistency with hypothesis. Moreover, both tenure and position level also demonstrated positive and significant relationship with work engagement. (Tenure: r = .14, p < .01;
Position Level: r = .21, p < .01). All variables’ number, means, standard deviations, reliability analysis and correlations were showed in Table 4.2., please refer to it for more details.
45
Table 4. 2.
Mean, Standard Deviations, Correlations, and Reliability (N=435)
Mean SD 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
1. Gender .43 .49
2. Position level 1.2 .55 -.09
3. Tenure 3.7 2.10 -.003 .35**
4. Acquiescent Silence 3.1 1.04 .03 .06 .18** (0.8)
5. Defensive Silence 3.2 .99 -.01 -.03 -.06 .61** (0.8)
6. Pro-social Silence 3.3 .95 -.03 .03 .06 .55** .70** (0.8)
7. Opportunistic Silence 2.8 .96 .09 .01 -.004 .60** .67** .57** (0.7)
8. Employee Well-Being 3.3 .73 -.04 .08 -.04 -.44** -.28** -.18** -.23** (0.9)
9. Mentoring Relationship 4.3 1.3 -.01 .06 -.06 -.36** -.19** -.14** -.16** .67** (0.9)
10. Work Engagement 4.3 1.2 -.02 .21** .14** -.23** -.19** -.07 -.15** .77** .65** (0.9) Note. Number in the brackets represent the Cronbach’s Alpha value of the variables. *p < .05 **p < .01 ***p < .001. Gender: 1 = Male, 0 = Female; Position level: 1 = General Staff, 2 = Line Manager, 3 = Middle Manager, 4 = High Manager ; Tenure: 1 = 0~1 year, 2 = 1~3 years, 3 = 3~5 years, 4 = 5~7 years, 5 = 7~10 years, 6 = 10~15 years, 7 = 15~20 years, 8 = 20~25 years, 9 = 25~30 years and 10 = above 30 years
46
Path Analysis in Structural Equation Modeling
Hypotheses were tested using path analysis in SEM (Gefen, Straub, & Boudreau, 2000) to examine the casual relationships and mediation effect among all the variables in this study. This approach was employed to create a path model and to analyze the structural relationship among the variables between different types of employee silences, employee well-being, and work engagement. Refer to Figure 4.1. for path model of work engagement as the dependent variable and employee well-being as the mediating variable.
Figure 4. 1. Path model of work engagement as the dependent variable and employee well-being as the mediating variable (N = 435).
Because the model is complex, all the relative fit indices are a little bit lower than the suggested value of above .90. The other fit indices of the model are within the acceptable range. Refer to the results summarized in Table 4.3.
47
Table 4. 3.
Summary of Goodness of Fit for Path Model of Work Engagement as the Dependent Variable and Employee Well-Being as the Mediating Variable (N = 435)
The model demonstrated that Acquiescent, Defensive, Pro-social and Opportunistic Silences had significant direct effects on employee well-being. Specifically, Acquiescent silence had a significant negative effect on employee well-being (path coefficient = -.63, p
< .001). Therefore, hypothesis 1a was confirmed. It was also found that Defensive Silence had a significant negative influence on employee well-being (path coefficient = -.40, p
< .05). Hence, hypothesis 1b was confirmed. Pro-social silence had a significant positive effect on employee well-being (path coefficient = .35, p < .05). Hence, hypothesis 1c was confirmed. Lastly, Opportunistic silence had a significant positive effect on employee well-being (path coefficient = .33, p < .05). Hence, hypothesis 1d was not confirmed. Refer to Table 4.3 for the results of direct, indirect and total effects of variables on work engagement.
Refer to Table 4.4 for the direct and indirect effects of studied variables on work engagement, which shows all types of silence have a significant direct effect on work engagement. Furthermore, indirect effect was calculated by multiplying the direct effect from different types of employee silences to employee wellbeing and the direct effect from employee well-being to work engagement.
Table 4. 4.
Direct and Indirect Effects of Variables on Work Engagement Acquiescent
48
Sobel Test
Sobel (1982) developed a formula to examine how the mediator carries the influence of the independent variable to the dependent variable, drawing upon the critical value of 1.96, to determine whether the results are supportive of mediation. Sobel test was used to determine whether significant indirect effects exist between Acquiescent, Defensive, Pro-social, Opportunistic Silences and work engagement (DV), when controlling for the mediator of employee well-being.
Results suggest that the relationship between acquiescent silence and work engagement is significantly mediated by employee well-being (z’ = -5.03, p < 0.01).
Secondly, defensive silence and work engagement is significantly mediated by employee well-being (z’ = -1.96, p < 0.05). Thirdly, the results indicated that employee well-being has a significant mediation effect between pro-social silence and work engagement (z’ = 2.69, p < 0.01). Lastly, the relationship between Opportunistic Silence and work engagement is also mediated by employee well-being (z’ = 2.27, p < 0.05). Furthermore, this research applied the variance accounted for (VAF) value (Latan & Noonan, 2017, pp.179) to test whether the mediator has a full or partial mediation effect. When the indirect effect divided by direct effect is over 80%, it represents a full mediation. After testing by VAF value, Acquiescent, Defensive and Opportunistic silences showed full mediation effects, however, Pro-social silence showed a partial mediation effect
In summary, Acquiescent, Defensive, Pro-social and Opportunistic Silences exerted a greater total effect to work engagement with indirect effects mediated through employee well-being. Therefore, it confirmed that there were mediation effects through employee well-being in the relationships between Acquiescent, Defensive, Pro-social, Opportunistic Silences and work engagement. Hence, hypotheses 3a ,3b ,3c and 3d were supported. Refer to Table 4.4. for the results of direct, indirect and total effects of variables on work engagement.
49
Hierarchical Regression
This analysis is used when there are two or more independent variables; and the purpose of this regression technique is applied to test causation (Gray & Kinnear, 2012). In this study, the researcher runs hierarchical regression to analyze the direction and effect between the variables of employee silence, employee well-being, work engagement, mentoring relationship, and control variables. Hierarchical regression investigates the results according to the influence of phenomena at different levels analysis by providing a conceptual and statistical mechanism (Hofmann, 1997). Another advantage is Hierarchical regression is able to investigate relationship within a certain hierarchical level and also investigate relationships across or between hierarchical levels at the same time (Bryk &
Raudenbush, 1992).
The regression result of all hypotheses was presented in Table 4.5. Model 1 and 2 demonstrated that Acquiescent and Defensive Silence were significantly and negatively related to work engagement (β = -0.26, p < 0.001; β = -0.18, p < 0.001), and also demonstrated that Opportunistic Silence was significantly and negatively related to work engagement (β = -0.16, p < 0.001). Accordingly, H2a, H2b and H2d were fully supported.
However, Model 3 demonstrated there was no significant relationship between Pro-social Silence and Work Engagement, so H2c was not supported.
This analysis was used to examine the moderating effect of mentoring relationship between different types of employee silences and work engagement. Control variables considered for this analysis were Tenure and Position Level. Refer to Table 4.6., the R2 increased from Model 1to Model 3, which all represented that mentoring relationship positively moderated the impacts of, Acquiescent, Defensive Silence, Prosocial Silence and Opportunistic Silence on work engagement. However, from the results showed in Model 3, which demonstrated that only the interaction of (OS X ME) has positively and significant (β = 0.22, p < 0.001) effect on work engagement. Refer to Figure 4.2. for further drawing the interaction plots to check the moderating effect.
50
Table 4. 5.
Result of Regression Analysis for Different Types of Employee Silences as Independent Variable (N = 435)
Notes: *p < 0.05. **p < 0.01. ***p < 0.001.
Work Engagement
Variables Controls
Position Level Tenure Main Effect
Acquiescent Silence Defensive Silence Prosocial Silence Opportunistic Silence
Model 1 Model 2 Model 3 Model 4
.108*** .184*** .112** .189***
.029* .049 .063 .057
-.259***
-.175***
-.078
-.155***
F 18.368*** 12.259*** 8.229 11.167***
R2 .113 .079 .054 .072
Adjusted R2 .107 .072 .048 .066
ΔR2 .065*** .030*** .006 .024***
51
Table 4. 6.
Result of Regression Analysis for Moderating Effect of Mentoring Relationship(N = 435)
Note. *p < 0.05. **p < 0.01. ***p < 0.001.
Work Engagement
Variables Model 1 Model 2 Model 3
Controls
Position Level .183*** .122** .120**
Tenure .073 .124** .125**
Main Effect
Acquiescent Silence .004 -.038
Defensive Silence -.082 -.062
Prosocial Silence .094 .105*
Opportunistic Silence -.054 -.058
Mentoring Relationship .639*** .639***
Interaction Effect
AS X ME -.030
DS X ME -.048
PS X ME -.083
OS X ME .216***
F 10.936*** 54.653*** 36.822***
R2 .048 .473 .489
Adjusted R2 .044 .476 .476
ΔR2 .048*** .424*** .017***
52
The interaction plot for the moderating effect of mentoring on the impacts of opportunistic silence and work engagement is presented in Figure 4.2. According to Aiken, West and Reno (1991) approach, first, the researcher recoded the moderator variable to a new categorical variable which stands for Low (1), Medium (2), High (3) mentoring. For example, low mentoring group means mentoring scores -1 SD, that is, everyone who scores below the number of “mean minus 1 SD”. On the other hand, high mentoring group means mentoring scores +1 SD, that is, everyone who scores above the number of “mean plus 1 SD”.
The low mentoring (solid line) group shows a strong negative relationship between opportunistic silence and work engagement, which means the group of employees who perceived weak mentoring relationship exhibit stronger negative relationship between opportunistic silence and work engagement. So, employees who had high opportunistic silence behavior will have low work engagement in low mentoring group. Conversely, the high mentoring (short-dash line) group has a weaker negative relationship between opportunistic silence and work engagement. That is, for the group of employees who perceived strong mentoring relationship, the negative effect of opportunistic silence on work engagement is almost non-existent. Therefore, employees who exhibit high opportunistic silence behavior can still have high work engagement in high mentoring group.
53
Figure 4. 2. Interaction plot for the moderating effect of mentoring relationship on opportunistic silence and work engagement.
High Middle Low
Mentoring Relationship Level of
Mentoring Relationship
54
Summary of Analysis Results
By investigating the influence of Acquiescent Silence, Defensive Silence, Prosocial Silence and Opportunistic Silence on work engagement. Also, whether employee well-being have mediating effect and mentoring relationship moderated the variables. This research aimed to provide more information for organization and researcher to understand silence situation in company. Additionally, this research aimed to further explore whether mentoring relationship would serve as a moderator on the influence of different types of employee silence on work engagement. Hypotheses were thereby generated to verify the relationships among variables, several statistical techniques were applied during the data analyzing process.
Firstly, the Pearson Correlation Analysis was performed for preliminary examining the correlations and directions between variables. The demographic variables as position level and tenure have positive and significant effect on work engagement. Secondly, Structural Equation Modeling was performed for examining the mediating effect of employee well-being between different types of employee silences and work engagement.
The result indicated that four different types of employee silences exerted a greater total effect to work engagement with indirect effects mediated through employee well-being.
Furthermore, the result indicated that Acquiescent Silence, Defensive Silence and Opportunistic Silence on employee well-being were significantly negative and Prosocial Silence were significantly positive. Thirdly, a Multiple Regression Analysis was performed for testing the hypotheses and examine the moderating effect of mentoring relationship. The result demonstrated that mentoring relationship significantly moderates the relationship between Opportunistic silences and work engagement. Hence, H1a, H1b, H1c, H2a, H2b, H2d, H3a, H3b, H3c, H3d, and H4d were supported. Summary of the results of hypotheses is shown in Table 4.7.
55
Table 4. 7.
Hypothesis Testing Results Summary
Hypothesis Results
Hypothesis 1a: Acquiescent Silence will have a negative effect on
employee well-being in organization. Supported Hypothesis 1b: Defensive Silence will have a negative effect on
employee well-being in organization. Supported Hypothesis 1c: Pro-social Silence will have a positive effect on
employee well-being in organization. Supported Hypothesis 1d: Opportunistic Silence will have a negative effect
on employee well-being in organization.
NOT Supported Hypothesis 2a: Acquiescent Silence will have a negative effect on
work engagement in organization. Supported
Hypothesis 2b: Defensive Silence will have a negative effect on
work engagement in organization. Supported
Hypothesis 2c: Pro-social Silence will have a positive effect on work engagement in organization.
NOT Supported
Hypothesis 2d: Opportunistic Silence will have a negative effect
on work engagement in organization. Supported Hypothesis 3a: Employee well-being has a mediating effect on the
relationship between Acquiescent Silence and work engagement.
Supported
Hypothesis 3b: Employee well-being has a mediating effect on the relationship between Defensive Silence and work engagement.
Supported
Hypothesis 3c: Employee well-being s has a mediating effect on the relationship between Pro-social Silence and work engagement.
Supported
56
Hypothesis 3d: Employee well-being has a mediating effect on the relationship between Opportunistic Silence and work engagement.
Supported
Hypothesis 4a: The mentoring relationship will weaken the relationship between Acquiescent Silence and work engagement.
NOT Supported
Hypothesis 4b: The mentoring relationship will weaken the relationship between Defensive Silence and work engagement.
NOT Supported
Hypothesis 4c: The mentoring relationship will weaken the relationship between Pro-social Silence and work engagement.
NOT Supported
Hypothesis 4d: The mentoring relationship will weaken the relationship between Opportunistic Silence and work engagement.
Supported
57
58