This study employs a quantitative approach; hence Hierarchical Linear Models and SPSS 22 was used for the data analysis. The researcher used descriptive statistics to examine sample profile, mean and standard deviation, Pearson correlation analysis for examining association between pairs of variables, HLM to test hypotheses 1 to 3 in cross level and hierarchical regression to test hypotheses 1 to 3 in individual level.
Descriptive Statistics and Correlation Analysis
Means, standard deviations, reliabilities, and inter correlations (Pearson’s correlation) of the variables are present in Table 4.1. As shown in the correlation analysis, organizational citizenship behavior (OCB) is negatively correlated with age in the organization (r=-.19, p<.05) which means younger employees are more likely to have organizational citizenship behavior than older employees. OCB is positively correlated with salary (r=0.28, p<.01) which means employees who have higher salary are more likely to perform more organizational citizenship behavior. OCB is positively correlated with social desirability (r=.19, p<.05), which means employees who have higher social desirability in organizations are more likely to have more organizational citizenship behavior. OCB is negatively correlated with the problem-solving team (r=-.17, p<.05) which means people who belong to the problem-solving teams are less likely to perform OCB in the organization. OCB is positively correlated with the self-management team (r=.22, p<.01), which means people who stay in self-self-management teams are more likely to have organizational citizenship behaviors.
Correlations were also used to examine direct and linear relationships between the independent variable, moderators and the outcome variable. As presented in research question one: Will Psychological Capital have an effect on Organizational Citizenship
65
Behavior? The correlation results between Psychological Capital and OCB shows a positive relationship (r=.69, p<.01), which means employees who have more positive Psychological Capital are more likely to perform more OCB. The second and the third questions raised in this study were as follows: Does a moderator effect of Team Identification exist on the relationship between Psychological Capital and Organizational Citizenship Behavior? Does a moderator effect of Team Cohesion exist on the relationship between Psychological Capital and Organizational Citizenship Behavior? The correlation result between Team identification and OCB shows a positive relationship (r=.66, p<.01), which means employees who have higher team identification tend to have more OCB. And the correlation result between Team cohesion and OCB shows a positive relationship (r=.58, p<.01), which means employees who have higher team cohesion also tend to have more OCB. Although the correlation results between team identification, team cohesion and OCB are promising, the research questions remain since correlation analysis cannot test the moderating effect. Next, the researcher used hierarchical linear and nonlinear modeling (HLM) analyses to test the moderating effect. The result of descriptive statistics and correlation analysis is shown in Table 4.1
66
Table 4.1.
Means, Standard Deviations, Reliabilities and Correlation among Study Variables
Note. Number in the brackets represents the Cronbach’s Alpha value. N=175
*p<.05. **p<.01.
Mean SD 1 2 3 4 5 6 7 8 9 10 11 12 13
1.Gender (male) .65 .48 2. Birth 34.71 6.72 .03
3. Company Tenure 45.57 32.18 .12 .41**
4. Position Tenure 43.29 32.96 .04 .43** .85**
5. Team Size 3.5 .90 -.05 .01 -.03 .03
6. Salary 2.7 .99 .13 .08 .46** .38** -.08 7. Social desirability 5.01 1.35 .00 .00 -.14 -.09 .06 -.02 8. Problem solving team .11 .32 .00 .07 .13 .11 .06 -.11 -.14
9. Self-management team .62 .49 -.06 .10 .03 .06 -.04 .16* .07 -.46**
10. PsyCap 4.74 .45 .08 -.14 .13 .08 -.00 .30** .26** -.19* .15* (.87)
11. Team Identification 3.80 .49 .08 -.31** .06 -.05 -.16* .33** .18* -.23** .13 .66** (.81)
12. Team Cohesion 4.96 1.03 .02 -.28** -.06 -.11 -.20** .28** .11 -.29** .16* .60** .78** (.93)
13. OCB 4.03 .45 .03 -.19* .07 .02 .03 .28** .19* -.17* .22** .69** .66** .58** (.82)
67
Hierarchical Linear Model Analysis
Hierarchical relationship occurred when variable at one level are influenced by variables at another level. HLM investigates the results according to the influence of phenomena at different levels analysis by providing a conceptual and statistical mechanism (Hofmann, 1997). HLM allows to analyze multilevel model by integrating individual and higher level data, therefore, HLM is able to investigate multilevel phenomena better. Another advantage is HLM 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).
Before HLM analysis, the researcher examined if the individual level data was appropriate to be aggregated into group level data. Therefore, the researcher needed to examine firstly the consistency within the group and secondly variation between groups.
There are three criteria to examine the consistency within the group and variation between groups: (1) rwg > .70 (Zohar, 2000), (2) ICC (1) > 0.12, (3) ICC (2) >.70 (Lin, 2005).
The rwg is one of the interrater agreement indices that is used to examine the homogeneity of identification and cohesion in work groups. rwg equal or greater than .70 may be considered high enough agreement to establish an interrater agreement (James, Demaree, & Wolf, 1984). In this research, Team identification rwg mean was 0.93, minimum=0.72, maximum=0.99. Team Cohesion rwg mean was 0.92, minimum=0.78, maximum=0.98. Both passed rwg test.
The purpose of ICC (1) is investigating the extent to which individuals are dependent on a team or organizations. Each subject is assessed by a different set of randomly selected respondents. ICC (1)＝τ00／（σ2＋τ00）, τ00 is the between-
68
group variance and σ2 is the within- group variance. According to Hox (2002) the effect of team should be concerned when ICC (1) is greater than 0.12. In this study, ICC (1) is 0.35; the result is acceptable.
ICC (2) is a measure for testing the stability of group means. Each respondent is viewed as representing a larger population. ICC（2）＝k(ICC（1）)／1＋（k－1）
(ICC（1）) According to Lin (2005), the value of ICC (2) greater than 0.70 is acceptable.
In this research, ICC (2) is 0.64, which is a bit lower than the criteria, thus becoming a limitation of this study.
Based on the rwg, ICC (1), and ICC (2) result shown above, the researcher aggregated the individual-level data into team-level data and tested the hypothesis in HLM.
Psychological Capital as Independent Variable
At first, the researcher examined H1: Psychological capital is positively related to employee citizenship behavior. Because of Psychological Capital and Organizational Citizenship Behaviors both are individual level, therefore, using HLM or SPSS both are acceptable. The following model provides a test of hypothesis 1 in HLM.
Level 1: Yij＝β0j＋β1j(PsyCap)+γij
The result showed that Psychological capital has a positive effect (γ10=0.63, SE=0.06, T-ratio=9.92, p<.001) on organizational citizenship behaviors, therefore, H1 is supported.
69
Team Identification as Moderator
Next, the researcher used HLM to tested Hypothesis 2: Team identification moderates the relationship between psychological capital and employee citizenship behavior. The hypothesis argues if team identification is stronger the relationship between psychological capital and organizational citizenship behavior will be stronger.
This model provides a test of hypothesis 2 in HLM:
Level 1: Yij＝β0j＋β1j(PsyCap)+γij
Level 2: β0j＝γ00+ γ01(TID)+U0j
β1j＝γ10+ γ11(TID)+U1j
Where:
Yij = Organizational Citizenship Behavior γ00 = Level 2 intercept
γ01 = Level 2 slope γ10 = Level 2 intercept γ11 = Level 2 slope
Variance (U0j) = τ00= residual intercept variance Variance (U1j) = τ11 = residual slope variance
The value of γ11 provides a test of team identification as a moderator; if the result is significant it means team identification has a moderator effect. As the result, γ11 is not significant (𝛾 11 =0.07, SE=0.17, T-ratio=0.42, n.s.), suggesting that Team Identification does not moderate the effect of psychology capital on organizational
70
citizenship behaviors. Therefore, Hypothesis 2 is rejected. However, the result suggests that in team level team identification has a direct effect on OCB (γ01 =0.41, SE=0.08, T-ratio=5.47, p<0.001).
Team Cohesion as Moderator
HLM is also used to tested Hypothesis 3: Team cohesion moderates the relationship between psychological capital and employee citizenship behavior. This hypothesis argues that if team cohesion is stronger the relationship between psychological capital and organizational citizenship behavior will be stronger. This model provides a test of hypothesis 3 in HLM:
Level 1: Yij＝β0j＋β1j(PsyCap)+γ Level 2: β0j＝γ00+ γ01(TCO)+U0j
β1j＝γ10+ γ11(TCO)+U1j
Where:
Yij = Organizational Citizenship Behavior Yij = Organizational Citizenship Behavior γ00 = Level 2 intercept
γ01 = Level 2 slope γ10 = Level 2 intercept γ11 = Level 2 slope
Variance (U0j) = τ00= residual intercept variance Variance (U1j) = τ11 = residual slope variance
71
As the result, γ11 is not significant (𝛾11 =0.06, SE=0.05, T-ratio=1.09, n.s.), suggesting that team cohesion does not moderate the effect of Psychology Capital on Organizational Citizenship Behaviors. Therefore, Hypothesis 3 is rejected. However, the result suggests that in team level team cohesion has a direct effect on OCB (γ01
=0.19, SE=0.03, T-ratio=5.56, P<0.001)
Hierarchical Regression Analysis
In the HLM analysis the researcher found that psychological capital has a positive relationship with organizational citizenship behavior, therefore, hypothesis 1 was confirmed. However, when testing Hypothesis 2 and 3 in HLM, the researcher found that team identification and team cohesion, as team-level variables, have more a direct effect on OCB rather than moderating effect. Therefore, Hypothesis 2 and 3 was rejected in HLM analysis. Next, the researcher uses regression in SPSS to examine if real estate agents’ perceptions of team identification and cohesion have moderator effects at the individual level. In SPSS the researcher uses hierarchical regression analyses to test the moderating effect.
Team Identification as Moderator.
The researcher used hierarchical regression to test Hypothesis 1and 2. In step 1 of the regression analysis, the researcher entered the control variables, including gender, age, education, organization tenure, problem solving team dummy, Self-management team dummy and social desirability. In step 2, Psychology Capital was entered to estimate its main effect on OCB. In step 3, Team identification was entered to estimate its main effect on OCB. In step 4, the control variable self-management team was a
72
significant predictor Organizational Citizenship Behaviors (𝛽= 0.13, p<0.5), suggesting that self-management team has positive effect on OCB. Team identification was entered to estimate its main effect on OCB. The finding was that the main effect of team identification on OCB was significant (β=.46, p<.001). The researcher mean-centered the Psychological Capital and Team identification before multiplying the two variables to create an interaction term. The interaction term of PsyCap and Team identification was not a significant predictor of OCB (β= -0.05, n.s.), suggesting that Team Identification does not moderate the effect of Psychology Capital on Organizational Citizenship Behaviors. Therefore, Hypothesis 2 was rejected. The result of Team Identification as moderator shows in Table 4.2.
Table 4. 2.
Regression Analysis: Team Identification as Moderator (N=104)
Predictors Organizational Citizenship Behaviors
Model 1 Model2 Model 3 Model 4
Gender 0.02 -0.02 -0.04 -0.03
Age -0.28*** -0.10 0.02 0.02
Education 0.05 0.07 0.05 0.05
Organization
Tenure 0.21** 0.01 -0.04 -0.02
Problem-solving
team -0.06 0.03 0.08 0.09
Self-management
team 0.20* 0.13* 0.12* 0.13*
Social desirability 0.20** 0.01 -0.01 0.00
(continued)
73
Table 4.2. (continued)
Predictors Organizational Citizenship Behaviors
Model1 Model2 Model3 Model4
PsyCap 0.70*** 0.43*** 0.42***
Team identification 0.47*** 0.46***
PsyCap × Team identification -0.05
𝐹 4.83 27.07 37.69 33.95
Adj. 𝑅2 0.13 0.55 0.66 0.65
∆ 𝑅2 0.17*** 0.40*** 0.11*** 0.00
Note. *p<.05. **p<.01.
Team Cohesion as Moderator.
The researcher also used hierarchical regression to test Hypothesis 3. In step 1 of the regression analysis, the researcher entered the control variables, including gender, birth, education, organization tenure, problem solving team dummy, Self-management team dummy and social desirability. In step 2, Psychology Capital was entered to estimate its main effect on OCB. In step 3, Team cohesion was entered to estimate its main effect on OCB. In step 4, the control variable self-management team was a significant predictor Organizational Citizenship Behaviors (β= 0.12, p<0.5), suggesting that self-management team has positive effect on OCB. Team cohesion was entered to estimate its main effect on OCB. The finding was that the main effect of team cohesion on OCB was significant (β=.26, p<.001). The researcher mean-centered the Psychological Capital and Team cohesion before multiplying the two variables to create an interaction term. The interaction term of PsyCap and Team cohesion was not a significant predictor of OCB (β= -0.02, n.s.), suggesting that Team Cohesion does not moderate the effect of Psychology Capital on Organizational Citizenship Behaviors.
74
Therefore, Hypothesis 3 was rejected. The result of team cohesion as moderator shows in Table 4.3.
Table 4. 3.
Regression Analysis: Team Cohesion as Moderator (N=104)
Predictors Organizational Citizenship Behaviors
Model 1 Model2 Model 3 Model 4
Gender 0.02 -0.02 -0.02 -0.02
Birth -0.28*** -0.10 -0.06 -0.06
Education 0.05 0.07 0.06 0.06
Organization
Tenure 0.21** 0.01 0.03 0.03
Problem solving
team -0.06 0.03 0.08 0.08
Self-management
team 0.20* 0.13* 0.12* 0.12*
Social desirability 0.20** 0.01 0.03 0.03
PsyCap 0.70*** 0.55*** 0.54
Team cohesion 0.26*** 0.26***
PsyCap × Team cohesion -0.02
𝐹 4.83 27.07 28.06 25.13
Adj. 𝑅2 0.13 0.55 0.58 0.58
∆ 𝑅2 0.17*** 0.40*** 0.04*** 0.00
Note. *p<.05. **<.01
75
Four Dimensions as Dependent Variable in HLM
Next, the researcher used HLM to explore Psychological Capital’s 4 dimension, including self-efficiency, hope, resilience and optimism. Examined if Team identification moderates the relationship between each dimension (self-efficiency, hope, resilience and optimism) and employee citizenship behavior. This model provides a test in HLM:
Level 1: Yij＝β0j＋β1j(PsyCap)+γij
Level 2: β0j＝γ00+ γ01(TID)+U0j
β1j＝γ10+ γ11(TID)+U1j
Where:
Yij =self-efficiency, hope, resilience and optimism γ00 = Level 2 intercept
γ01 = Level 2 slope γ10 = Level 2 intercept γ11 = Level 2 slope
Variance (U0j) = τ00= residual intercept variance Variance (U1j) = τ11 = residual slope variance
The value of γ11 provides a test of team identification as a moderator; if the result is significant it means team identification has a moderator effect. As the result, Self-efficiency’s γ11 is not significant (𝛾11 =0.22, SE=0.19, T-ratio=1.20, n.s.), Hope’s γ
11 is not significant (γ11 =-0.22, SE=0.14, T-ratio=-1.57, n.s.), Resilience’s γ11 is not significant (γ11 =-0.23, SE=0.18, T-ratio=-1.29, n.s.), suggesting that Team Identification does not moderate the effect of these three dimension on organizational
76
citizenship behaviors.Optimism’s γ11 is not significant (γ11 =-0.49, SE=0.17, T-ratio=--2.87, p<0.1), suggesting that Team Identification moderate the effect of these three dimension on organizational citizenship behaviors.
Table 4. 4.
Team Identification as Moderator in HLM
Variable γ11 SE T-ratio Significance
Self-efficiency 0.22 0.19 1.20 n.s.
Hope -0.22 0.14 -1.57 n.s.
Resilience -0.23 0.18 -1.29 n.s.
Optimism -0.49 0.17 -2.87 0.007**
Note. **p<.01
The researcher also used HLM to explore Psychological Capital’s 4 dimension, including self-efficiency, hope, resilience and optimism. Examined if Team cohesion moderates the relationship between each dimension (self-efficiency, hope, resilience and optimism) and employee citizenship behavior. This model provides a test in HLM:
Level 1: Yij＝β0j＋β1j(PsyCap)+γij
Level 2: β0j＝γ00+ γ01(TCO)+U0j
β1j＝γ10+ γ11(TCO)+U1j
Where:
Yij =self-efficiency, hope, resilience and optimism γ00 = Level 2 intercept
γ01 = Level 2 slope
77
γ10 = Level 2 intercept γ11 = Level 2 slope
Variance (U0j) = τ00= residual intercept variance Variance (U1j) = τ11 = residual slope variance
The value of γ11 provides a test of team cohesion as a moderator; if the result is significant it means team cohesion has a moderator effect. As the result, Self-efficiency’s γ11 is not significant (𝛾11 =0.04, SE=0.09, T-ratio=0.46, n.s), Hope’s γ11
is not significant (γ11 =-0.20, SE=0.11, T-ratio=-1.87, n.s), Resilience’s γ11 is not significant (γ12 =-0.23, SE=0.10, T-ratio=-1.17, n.s), suggesting that Team cohesion does not moderate the effect of these three dimension on organizational citizenship behaviors. Optimism’s γ11 is not significant (γ11 =-0.26, SE=0.09, T-ratio=--2.88, p<0.1), suggesting that Team cohesion moderate the effect of these three dimension on organizational citizenship behaviors.
Table 4.5.
Note. **p<.01
Modified Framework
As the research result, the researcher suggested that the research framework should add team identification and team cohesion into independent variables and examined the
Variable γ11 SE T-ratio Significance
Self-efficiency 0.04 0.09 0.46 n.s.
Hope -0.20 0.11 -1.87 n.s.
Resilience -0.12 0.10 -1.17 n.s.
Optimism -0.26 0.09 -2.88 0.006**
Team Cohesion as Moderator in HLM
78
importance of team identification and team cohesion effect on Organizational citizenship behaviors in real estate industry. The modified research framework is presented in Figure 4.1. and Figure 4.2.
Figure 4.1. Team identification and cohesion as independent variable.
Figure 4.2. Optimism as independent variable.
79