This chapter presented findings based on the study hypotheses. In the first part, the descriptive statistics analysis was presented. The second part included the result of correlation analysis among cultural intelligence (CQ), cross-cultural adjustment, psychological capital (PsyCap), and perceived supervisor support (PSS). The third parts showed the validity of the study. The fourth part focused on the findings of hypotheses testing by using structure equation model (SEM) and hierarchical regression analysis.
Descriptive Statistics
The demographic information include age, duration in Taiwan, overseas experience, marriage, educational attainment, and language ability. Owing to the characteristics of the industry, all of the 538 participants were female. Most of them were single (73.4%) and aged from 26 to 30 years old (41.3%). The participants had stayed in Taiwan for three to six years principally (42.5%). As for the overseas and working experience, a great percentage of them did not work or live in other countries aside from their home country (91.6%); however, most of them have work experience in electronic industry before coming to Taiwan. Regarding their education background, over 66.5%
participants had bachelor degree and over half of them (56.7%) rated their English ability (both listen-speaking and reading-writing) at the level of the good. On the contrary, a great number of participants rated their Chinese ability at the level of the poor (60%). The frequency and percentage of the demographic information are summarized in Table 4.1.
46
Table 4.1.
Descriptive Statistics (n=538)
Frequency Percentage Item Frequency Percentage
1. Age 5. Marriage
47
Correlation Analysis
In order to understand the relationship between two variables, the Pearson correlation analysis was performed. The means, standard deviations, reliabilities, and correlations among all of the variables are presented in Table 4.2. The major variables in this study showed significant correlation coefficients. CQ was positively correlated to PsyCap (r=.48, p<.001), cross-cultural adjustment (r=.54, p<.001), and the three sub-dimensions of cross-cultural adjustment (r=.47, r=.46, r=.48, p<.001). Also, CQ was moderately correlated with PSS (r=.30, p<.001). Cross-cultural adjustment was positively correlated with PsyCap (r=.43, p<.001) and PSS (r=.42, p<.001).
48
Table 4.2.
Mean, Standard Deviations, Correlations, and Reliability (n=538)
Mean S.D. 1 2 3 4 5 6 7 8 9 10 11 12
1. Age (year) 30.48 4.47
2. Overseas experience (month) 2.36 11.12 .12**
3. Duration in Taiwan (month) 46.04 31.80 .54*** .06
4. English ability 3.93 0.60 -.12*** -.00 .02
5. Chinese ability 2.15 0.50 -.00 -.02 .08 .21***
6. CQ 5.00 0.69 -.04 -.01 .01 .19*** .13*** (.92)
7. Cross-cultural adjustment 5.32 0.77 -.01 -.01 .06 .14*** .11** .54*** (.94)
8. General adjustment 5.36 0.83 .02 .01 .07 .15*** .08 .47*** .94*** (.91)
9. Interaction adjustment 5.07 0.93 -.02 -.01 .08 .10* .15*** .46*** .87*** .71*** (.91)
10. Work adjustment 5.54 0.91 -.06 -.06 -.03 .11* .07 .48*** .78*** .61*** .56 *** (.89)
11. Psychological capital 3.61 0.36 .02 .04 .04 .21*** .07*** .48*** .43*** .38*** .34*** .43*** (.87)
12. Perceived supervisor support 4.66 1.15 -.10 -.02 -.09* .10* .12*** .30*** .42*** .37*** .39*** .34*** .26*** (.84) Notes. Numbers in parentheses represent Cronbach’s alpha value. ***p < .001 **p < .01 *p < .05
49
Confirmatory Factor Analysis (CFA)
Confirmatory factor analysis (CFA) was performed to evaluate the distinctiveness of the measures by using AMOS 18.0. Since the chi-square test is sensitive to sample size, the overall model fit was also examined by other fit indices, including root-mean-square error of approximation (RMSEA), Tucker–Lewis non-normed index (TLI), incremental fit index (IFI), and the comparative fit index (CFI). When RMSEA is below .08 (Browne & Cudeck, 1993) and TLI, IFI and CFI scores are above .90 (Byrne, 1998) presents good model fit index.
The practice of item parceling was adopted as conducting confirmatory factor analyses on cross-cultural adjustment and PsyCap in order to reduce the measurement error and to increase the stability of the parameter estimates (Bagozzi & Edwards, 1998). However, a limitation of the data analysis occurred in regards of the identification issue. As for cross-cultural adjustment, the number of tree parameters exactly equals to the number of known values that caused zero degree of freedom, which made it a perfect just-identified model. Thus, there was no need to assess the CFA of cross-cultural adjustment. The original 24 items of PsyCap was divided into four parcels. The chi-square of PsyCap was 1.62, while the other fit indexes were:
RMSEA=0, CFI=1, NFI=1, IFI=1, TLI=1, GFI=1. CQ was tested by 20 items and the fit indexes were presented as follows: Chi-square (χ2)=451.84, RMSEA=.06, CFI=.95, NFI=.93, IFI=.95, TLI=.94, GFI=.92. PSS with four items also showed expected fit indexes; the chi-square (χ2) was 5.93, while the other fit indexes were: RMSEA=.06, CFI=.1, NFI=.99, IFI=.1, TLI=.99, GFI=.99. The variables in this study all presented good fit indexes and it confirmed the validity of the study.
Other than evaluating the validity of each variable, the validity of the full model was tested. The validity of the 27 constructs (three factors of cross-cultural adjustment, four factors of PsyCap, 20 factors of CQ, and four factors of PSS) presented
50
acceptable fit: Chi-square (χ2)=1149.29, RMSEA=.06, CFI=.91, NFI=.87, IFI=.92, TLI=.90, GFI=.88. Also, the alternative two-factor and one-factor model was conducted to show the discriminant validity among the multidimensional constructs.
The two-factor model reveals the unacceptable fit indices (χ2=3534.74, RMSEA=.14, CFI=.60, NFI=.58, IFI=.60, TLI=.56, GFI=.59). And the one-factor model also results in poor fit indices (χ2=3879.17, RMSEA=.14, CFI=.56, NFI=.54, IFI=.56, TLI=.52, GFI=.57). The chi-square differences indicated that the three-factor model was superior to the two-factor and the one-factor models.
Table 4.3.
Results of Confirmatory Factor Analysis (n=538)
Notes. RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index; NFI = Normed Fit Index; IFI = Incremental Fit Index; TLI = Taker–Lewis Index, GFI = Goodness of Fit Index
Because the data was self-reported and collected through the same questionnaire in the same time, Harman’s one-factor test (Podsakoff, MacKenzie, Lee,
& Podsakoff, 2003) was conducted to detect potential bias caused by common method variance (CMV). The assumption provided by Podsakoff et al. (2003) suggested that if a single factor emerged or one general factor accounts for the majority of all the dependent and independent variables means that the CMV problem presented in the study. In this study, all the 62 items were entered into the factor analysis and then the un-rotated factor solution was examined to determine the number of variances of the variables. The result reveals 13 factors with the eigenvalue greater than 1.0, which
51
and the result shows that the items did not load on a single factor.
Structure Equation Model (SEM)
Structure equation model (SEM) was designed to test the Hypothesis 1 to Hypothesis 4 in this study. As for testing the mediating effect of PsyCap, we used Baron and Kenny’s (1986) approach. Figure 4.1 and Figure 4.2 shows the result in details.
Hypothesis 1 predicted that CQ has a positive effect on cross-cultural adjustment. The SEM result showed that CQ had a significantly positive effect on cross-cultural adjustment (β=.682, p<.001). Specifically, CQ had a significant positive effect on three sub-dimensions of cross-cultural adjustment: general adjustment (β=.858, p<.001), interaction adjustment (β=.814, p<.001), and work adjustment (β=.713, p<.001). Therefore, Hypothesis 1-1, 1-2, and 1-3 were all supported and Hypothesis 1 was sustained.
Hypothesis 2 predicted that CQ has a positive effect on PsyCap. The SEM result presented that the structural coefficient β relating CQ to PsyCap was .584 (p<.001), providing support for Hypothesis 2.
Hypothesis 3 predicted that PsyCap has a positive effect on cross-cultural adjustment. The data indicates that PsyCap also had a positive effect on cross-cultural adjustment (β=.187, p<.01). Specifically, PsyCap had positive effects on the three sub-dimensions of cross-cultural adjustment: general adjustment (β=.863, p<.001), interaction adjustment (β=.807, p<.001), and work adjustment (β=.715, p<.001). The results revealed supports for Hypothesis 3-1, 3-2, and 3-3. Therefore, Hypothesis 3 was fully supported.
Meanwhile, Hypothesis 4 predicts that PsyCap serves as a mediating link between CQ and cross-cultural adjustment as well as the three sub-dimensions. The beta coefficient between CQ and cross-cultural adjustment decreased and remained
52
significant (from .682 to .577) when PsyCap was added as the mediator. Therefore, PsyCap played as a partial mediating link between CQ and cross-cultural adjustment.
The similar results were observed when testing the mediating effects of PsyCap on the relationship between CQ and general adjustment, CQ and interaction adjustment, and CQ and work adjustment. We can tell from Figure 4.1 and Figure 4.2 that the beta coefficients between CQ and general adjustment as well as between CQ and interaction adjustment were decreased and remained significant (.858 to .855; .814 to .810, respectively) after PsyCap was added as the mediator. These results indicated that PsyCap partially mediated the relationship between CQ and general adjustment as well as the relationship between CQ and interaction adjustment. The beta coefficient between CQ and work adjustment changed from significant to not significant after PsyCap was added. This indicates that PsyCap fully mediated the relationship between CQ and work adjustment. Therefore, Hypothesis 4.1 and Hypothesis 4.2 was partially supported, and Hypothesis 4.3 was fully supported.
53
Figure 4.1. SEM model with path coefficients
54
Figure 4.2. SEM mediating model with path coefficients
55
Hierarchical Regression Analysis
Hypothesis 5 predicted that PSS positively moderated the relationship between PsyCap and cross-cultural adjustment. The moderating effect of PSS was tested via hierarchical linear regression analysis. The three steps recommended by Baron and Kenny (1986) were adopted in this study.
In step one, the control variables such as age, overseas experience, duration in Taiwan, marriage, education attainment, work experience, English ability, and Chinese ability were entered. In step two, the independent variable and moderating variable were entered. In step three, the interactional variable was entered and we examined the beta coefficient and the significance of the interactional variable.
However, before calculating the interaction variable, PsyCap and PSS was centered by subtracting the mean from the original ones in order to reduce the multicollinearity problem (Aiken & West, 1991).
Table 4.4 to Table 4.7 summarize the regression result of testing Hypothesis 5 and Hypotheses 5-1 to 5-3. Model 1 presents that the interaction of PsyCap and PSS was significant when cross-cultural adjustment was entered as the dependent variable, and the beta coefficient was positive (β=.23, p<.01). Model 2 to Model 4 also present the similar results as the three sub-dimensions of adjustment were entered as the dependent variables (β=.22, p<.05 for general adjustment; β=.25, p<.05 for interaction adjustment; β=.21, p<.05 for work adjustment).
To interpret the moderating effect in detail, PSS was divided into high and low level group based on the mean. Figure 4.2 depicts the interactional graphs between PsyCap and PSS. As expected, PsyCap had a more positive effect on cross-cultural adjustment, general adjustment, interaction adjustment, and work adjustment for individuals with higher PSS; whereas, PsyCap had a more slight effect on cross-cultural adjustment as well as the three sub-dimensions for individuals with
56
lower PSS. To sum up, PSS showed significantly positive interactional effect between CQ and cross-cultural adjustment and its three-sub-dimensions. Hence, Hypothesis 5 and Hypotheses 5-1 to 5-3 were supported.
In conclusion, individuals who perceived higher level of supervisor support performed higher level of cross-cultural adjustment than those who perceived low level of supervisor support. The emotional support, information support, and instrumental support provided by the supervisors were helpful for individuals to overcome stress occurring in the adaptation process (Caligiuri & Lazarova, 2002). In addition, the other tangible or intangible support (e.g., knowledge sharing, entertainment, and welfare) reduced the uncertainty and confusion when individuals moved to an unfamiliar environment. Kraimer et al. (2001) also noted that individulas who received aids, rewards, and affirmation from the supervisors had more encouragement to take over the challenging tasks and to perform successful work adjustment.
57
Table 4.4.
Result of Regression Analysis of the Moderating Effect Between PsyCap and Cross-cultural Adjustment (n=538)
Variables
Model 1 Model 2 Model 3
β β β
Step 1: Controls
Age -.01 -.01 -.01
Overseas experience .00 .00 .00
Duration in Taiwan .00 .00 .00
Education attainment .00 -.02 -.02
Work experience -.09 -.11 -.10
English Ability .14 .02 .00
Chinese Ability .12 .04 .04
Step 2: Main Effect
Psychological capital .74*** .71***
Perceived supervisor
support .23*** .23***
Step 3: Interaction
PsyCap x Pss .23**
R2 .03 .30 .31
Adj. R2 .02 .29 .30
F 1.97* 21.24*** 20.33***
∆ R2 .27 .01
∆F 95.32*** 8.11***
Note. ***p < .001 **p < .01 *p < .05
58
Table 4.5.
Result of Regression Analysis of the Moderating Effect Between PsyCap and General Adjustment (n=538)
Variables
Model 1 Model 2 Model 3
β β β
Step 1: Controls
Age -.01 .00 .00
Overseas experience .00 .00 .00
Duration in Taiwan .00 .00 .00
Education attainment .02 .00 .00
Work experience -.04 -.06 -.05
English Ability .17 .07 .04
Chinese Ability .07 .00 .00
Step 2: Main Effect
Psychological capital .66*** .63***
Perceived supervisor support
.23*** .23***
Step 3: Interaction
PsyCap x Pss .22*
R2 .03 .24 .25
Adj. R2 .01 .22 .23
F 1.80 15.22*** 14.51***
∆ R2 .21 .01
∆F 67.00*** 5.89***
Note. ***p < .001 **p < .01 *p < .05
59
Table 4.6.
Result of Regression Analysis of the Moderating Effect Between PsyCap and Interaction Adjustment (n=538)
Variables
Model 1 Model 2 Model 3
β β β
Step 1: Controls
Age -.02 -.02 -.02
Overseas experience .00 .00 .00
Duration in Taiwan .00 .00 .00
Education attainment .00 -.03 -.02
Work experience -.13 -.15 -.14
English Ability .08 -.03 -.06
Chinese Ability .23 .15 .15
Step 2: Main Effect
Psychological capital .68*** .66***
Perceived supervisor support .27*** .27***
Step 3: Interaction
PsyCap x Pss .25*
R2 .04 .24 .25
Adj. R2 .02 .23 .24
F .26** 15.84*** 15.11***
∆ R2 .20 .01
∆F 66.18*** 6.16*
Note. ***p < .001 **p < .01 *p < .05
60
Table 4.7.
Result of Regression Analysis of the Moderating Effect Between PsyCap and Work Adjustment (n=538)
Variables
Model 1 Model 2 Model 3
β β β
Step 1: Controls
Age -.01 .00 .00
Overseas experience .00 -.01 -.01
Duration in Taiwan .00 .00 .00
Education attainment -.02 -.05 -.05
Work experience -.16 -.18 -.17
English Ability .13 .00 -.02
Chinese Ability .08 .01 .01
Step 2: Main Effect
Psychological capital .99*** .97***
Perceived supervisor support .18*** .18***
Step 3: Interaction
PsyCap x Pss .21*
R2 .02 .25 .26
Adj. R2 .00 .24 .24
F 1.27 16.61*** 15.64***
∆ R2 .23 .01
∆F 76.46*** 4.69*
Note. ***p < .001 **p < .01 *p < .05
61
(continued) Figure 4.3. Interaction plots for the moderating effects of perceived supervisor support
Table 4.8.
Result of Hypotheses Testing
Hypotheses Result
H 1. Cultural intelligence has a positive effect on cross-cultural adjustment.
Supported
H 1-1. Cultural intelligence has a positive effect on general adjustment.
Supported
62
(continued) Table 4.6. (continued)
H 1-2. Cultural intelligence has a positive effect on interaction adjustment.
Supported
H 1-3. Cultural intelligence has a positive effect on work adjustment .
Supported
H 2. Cultural intelligence has a positive effect on psychological capital.
Supported
H 3. Psychological capital has a positive effect on cross-cultural adjustment.
Supported
H 3-1. Psychological capital has a positive effect on general adjustment.
Supported
H 3-2. Psychological capital has a positive effect on interaction adjustment.
Supported
H 3-3. Psychological capital has a positive effect on work adjustment.
Supported
H 4. Psychological capital serves as a mediator between cultural intelligence and cross-cultural adjustment.
Supported
H 4-1. Psychological capital serves as a mediator between cultural intelligence and general adjustment.
Partially supported H 4-2. Psychological capital serves as a mediator between cultural
intelligence and interaction adjustment.
Partially supported H 4-3. Psychological capital serves as a mediator between cultural
intelligence and work adjustment.
Supported
H 5. Perceived supervisor support positively moderates the relationship between psychological capital and cross-cultural adjustment.
Supported
63
Table 4.6. (continued)
H 5-1. Perceived supervisor support positively moderates the relationship between psychological capital and general adjustment.
Supported
H 5-2. Perceived supervisor support positively moderates the relationship between psychological capital and interaction adjustment.
Supported
H 5-3. Perceived supervisor support positively moderates the relationship between psychological capital and work adjustment.
Supported
64
65