This chapter presents the findings regarding research hypotheses of this study. The first part includes the results of descriptive statistical analysis and the characteristic analysis of each variable by using one-way ANOVA and t-test (hypothesis 1 to 3). The second part includes the results of correlations among HR practices, ECCC, and affective commitment (hypothesis 4), the results of multiple regression analysis to see if affective commitment will be influenced by HR practices and ECCC (hypothesis 5, 6), and the results of moderated regression to understand whether the relationship between HR practices and affective commitment will be moderated by ECCC (hypothesis 7).
General Characteristics of the Sample
Descriptive statistic analysis of sample
Nine items of demographic information including nationality, gender, marital status, age, education, year-in-Taiwan, monthly salary, religion and vocation were surveyed. The results are shown in Table 4.1.
To make the results more comparable, the groups over age 31 are combined into one, which includes 27 persons (11.84%). Similarly, since most of the migrant workers are senior high school graduates or have bachelor’s degree, the other groups are combined together into one (8.77%). Besides, the groups earn over NTD30,000 are combined as well, which occupied 2.19%.
All the respondents are manufacturing workers in Taiwan. For nationality, 132 respondents (57.89%) working for A company are from Philippines, while the rest 96 (42.11%) working for B company are from Indonesia. For gender, 42.11% of the respondents are females, all coming from Philippines, while 57.89% are males (32 from Philippines, 96 from Indonesia). For marital status, most of them are single (82.89%), 13.6% are married. As for age distribution, 87.29% are between age 21 to age 30. For education level, 55.26% have bachelor degree, while 35.09% graduated from senior high school. The year-in-Taiwan of respondents is distributed equally, 64.48% worked one to four years in Taiwan. For monthly salary, 71.49% of the respondents earn NTD15,001 to NTD25,000, 15.35% earn less than NTD15,000, while 13.15% earn more than NTD 25,000. The religion includes Buddhist (11.40%), Muslim (27.63%), Christian (15.35%), Catholic (42.11%) and other (3.51%).
Table 4.1 Demographic characteristics of sample
Item Description Frequency Percent
Company A 132 57.89
B 96 42.11
Nationality Philippines 132 57.89
Indonesia 96 42.11
Vocation Manufacturing worker 228 100.00
T-test and one-way ANOVA analysis of sample
HR practices
According to table 4.2, the results show that there is a significant difference of migrant worker’s perception scores of training, performance appraisal and reward between different nationality and gender groups. For nationality, the training, performance and reward scores of Philippine workers are significantly lower than Indonesian ones (p<.001). For gender, the training, performance appraisal and reward scores of male workers are significantly higher than female (p<.001). As for benefit, there is no significant difference between Philippine/Indonesian groups nor male/female groups. Besides, the marital status of migrant workers did not make a significant difference on any item of HR practices.
For the items of age and year-in-Taiwan, there is no significant difference on any scores of HR practices, which means different age and year-in-Taiwan groups did not make a difference on training, performance appraisal, reward and benefit scores, see table 4.2.
According to table 4.2, the scores of training, performance appraisal, and reward are significantly different among salary groups. Among them, migrant workers with salary between NTD 25,000 ~ NTD 30,000 have higher scores on performance appraisal and reward than those earn between NTD 15,000 ~ NTD 20,000 per month. Among differing education-level workers, there is a significant difference on scores of training, performance appraisal and reward. Senior high school workers score higher on training, performance appraisal and reward than bachelor’s degree ones. However, as mentioned, most of the high school workers are coming from Indonesia, while most of the Philippines have bachelor’s degrees.
As for religion, there is a significant difference on scores of training, performance appraisal, reward and benefit among different religious groups. Among them, Buddhists and Muslims score significantly higher than Christians on training and performance appraisal;
Buddhists score significantly higher than Christian and Catholics on reward; Muslims score significantly higher than Catholics on reward as well. But most of the Muslims and Buddhists are from Indonesian, while the most of the Christians and Catholics are Filipinos.
Table 4.2 T-test and one-way ANOVA analysis on HR practices
Training Performance Reward Benefit Nationality
Gender Male Female Male Female Male Female Male Female Mean 61.81 58.20 32.54 28.31 20.79 18.00 26.03 25.32
t-value 3.602*** 5.551*** 6.515*** 1.224
Marital s. Married Single Married Single Married Single Married Single Mean 60.42 60.07 32.29 30.33 24.97 25.83 20.29 19.43
t-value 0.234 1.674 -1.015 1.349
d: 25001~30000 and e: >30001
(b) Three educational levels a: Senior high school, b: Bachelor and c: Other (c) Five religion groups a: Buddhist, b: Muslim, c: Christian, d: Catholic and
e: Other
** p<.05 *** p<.001
ECCC
According to table 4.3, the results show that there is no significant difference on ECCC scores among any demographic groups, since all p-values do not achieve the standards of significance.
Affective commitment
For nationality, the Indonesian workers have a higher scores on affective commitment than the Philippine ones (p<.001). For gender, the male workers score higher on affective commitment than the female ones (p<.001). As for marital status, there is no significant difference on affective commitment between married and single workers, see table 4.3.
From the aspect of age, workers above age 31 have a significantly higher score on affective commitment than the other two groups – age below 25 and between 26~30. As for year-in-Taiwan, workers stayed in Taiwan for more than 6 years have a higher score on affective commitment than all the other groups, see table 4.3.
Table 4.3 T-test and one-way ANOVA analysis on ECCC and affective commitment ECCC Affective commitment
Marital status Married Single Married Single
Mean 21.41 21.04 28.71 27.81
Scheffé e>a, e>b, e>c, e>d
(a) Three age groups a: < 25, b: 26~30, and c > 31 years old (b) Five year-in-TW groups a: < 1, b: 1~2, c: 2~4, d:4~6 and e: > 6 years
** p<.05 *** p<.001
(table continues)
Table 4.3 (continued)
ECCC Affective commitment Salary (c)
F-value 1.523 5.029***
Scheffé c>b
Education (d)
F-value 0.169 15.330***
Scheffé a>b, c>b
Religion (e)
F-value 1.928 6.069***
Scheffé b>d
(a) Three age groups a: < 25, b: 26~30, and c > 31 years old (b) Five year-in-TW groups a: < 1, b: 1~2, c: 2~4, d:4~6 and e: > 6 years (c) Five salary groups a: <15000, b: 15001~20000, c: 20001~25000,
d: 25001~30000 and e: >30001
(d) Three educational levels a: Senior high school, b: Bachelor and c: Other
(e) Five religious groups a: Buddhist, b: Muslim, c: Christian, d: Catholic and e: Other
** p<.05 *** p<.001
Among different salary groups, there is a significant difference on affective commitment.
Workers earned between NTD 20,001~NTD 25,000 have a higher affective commitment scores than those earned between NTD 15,001~NT 20,000. For educational levels, workers with bachelor’s degree have a lower score than the other two groups (p<.001). For religion, there is a significant different score on affective commitment among different religious groups (p<.001). Muslims score significantly higher than Catholic on affective commitment, as shown in table 4.3.
All in all, different nationality, gender, salary and educational groups have significantly different scores on training, performance appraisal, reward and affective commitment.
Different religious groups have significant differences on four HR practices, ECCC and affective commitment, as shown in table 4.4.
Table 4.4 Summary of sample characteristics analysis
Training PA Reward Benefit ECCC AC
Nationality ☆ ☆ ☆ ☆
Gender ☆ ☆ ☆ ☆
Marital status
Age ☆
Year-in-TW ☆
Salary ☆ ☆ ☆ ☆
Education ☆ ☆ ☆ ☆
Religion ☆ ☆ ☆ ☆ ☆ ☆
☆ refers to significant difference among different demographic groups.
Further exploration
There are two reasons to further explore the data: (1) the results shown in table 4.4 are quite similar among nationality, gender, salary and educational groups, and (2) most of the Indonesian respondents are male, with high school degree and believe in Muslim and Buddhist; while the majority of the Philippine respondents are female, with bachelor’s degree and believe in Christian and Catholic, so the overlap among nationality, gender, education and religion is obvious. Thus, it is unapparent what the most important factors are. By Chi-square test of independence, it is ensured that nationality is significantly not independent of gender (p<.001), educational level (p<.001) and religion (p<.001) of the sample, as shown in table 4.5. Thus, the data is being further investigated by splitting into different demographic groups and using one-way ANOVA to compare the results.
Table 4.5 Chi-square (χ2) test of independence on demographic variables
Gender Salary Education Religion χ2=120.595 χ2=53.745 χ2=188.740 χ2=204.997
Nationality
p=.000 p=.000 p=.000 p=.000
As shown in table 4.6, among male workers, there is no significant difference of HR practices and ECCC scores between Philippine and Indonesian males, while the affective commitment scores of Indonesian males are significantly higher than the Philippine males.
Among Philippine workers, the performance and reward scores of Philippine males are significantly higher than Philippine females at 0.05 and 0.01 level.
As a whole, migrant workers with salary between NTD 25,000 ~ NTD 30,000 have higher scores on performance appraisal and reward than those earn between NTD 15,000 ~ NTD 20,000 per month. Workers earned between NTD 20,001 ~ NTD 25,000 have a higher affective commitment scores than those earned between NTD 15,001 ~ NTD 20,000, see table 4.2 and 4.3. While investigating Indonesian and Philippines separately, there is no significant difference in different Philippine salary groups. But among Indonesian workers, those earned between NTD 20,001 ~ NTD 25,000 have a higher affective commitment scores than those earned between NTD 15,001 ~ NTD 20,000, the same with the entire sample. This is reasonable. Since those who earned more are happier about their salary, they enjoy discussing their organizations with people and committing affectively to the organization accordingly.
Besides, those earned below NTD 15,000 have a higher perception on reward scores than those earned between NTD 15,001 ~ NTD 20,000. This could be explained: since the purposes of all migrant workers coming to Taiwan are making money and they know their salary in Taiwan will be higher than in their own countries, thus even if the salary of migrant workers within 1 year experience must be deducted from their agencies, they still think the reward of their company is quite good. However, while the migrant workers worked more than 1 year, their salary will no longer be taken off, but at the same time, they are more experienced and knowing about the situations in Taiwan. Thus, even though, their salaries become higher than before, Indonesians become unsatisfied with the reward system instead.
However, Philippines worker is another case. Because of their nationality, they are more acquainted with their rights, duties and obligations before working in Taiwan. They know exactly what the salary mechanism is, so the change on salary will not make their perceptions different.
Although table 4.2 and 4.3 shows that, among different education-level workers, there are significant differences on the scores of training, performance appraisal, rewards, and affective commitment (college > high school). While investigating Indonesian and Philippines separately, it is found that there is no significant difference among Philippine and
Indonesian education-level workers, see table 4.6. Thus, this determines that the differences in scores are from nationality, instead of education level in this study.
Table 4.6 One-way ANOVA among demographic groups
Training PA Reward Benefit ECCC AC
Male PHL <IDN
Nationality
F-value -2.332 -3.535 -.931 -1.520 -.214 -9.698**
PHL M >F M >F
Gender
F-value 2.013 5.447** 13.648*** 2.598 2.674 .189 PHL
F-value .980 1.200. .868 2.058 .967 .032
IDN a>b c>b
Salary (a)
F-value 1.180 2.388 3.096** .347 1.722 3.252**
PHL
F-value 1.148 1.684 .395 .862 .105 2.266
Education IDN
F-value .487 1.963 .989 1.751 .615 .809
PHL
F-value 3.856** .557 .898 1.150 .270 .379
Religion (b) IDN
F-value 4.691** 2.894** 8.080*** 5.009** 4.370** 2.213
PHL a >d,e
F-value .685 .744 .290 1.497 .430 3.316**
YIT (c) IDN
F-value .362 .093 .156 .950 .697 1.247
(a) Five salary groups a: <15000, b: 15001~20000, c: 20001~25000, d: 25001~30000 and e: >30001
(b) Five religious groups a: Buddhist, b: Muslim, c: Christian, d: Catholic and e: Other (c) Five year-in-TW groups a:< 1, b: 1~2, c: 2~4, d:4~6 and e: > 6 years
** p<.05 *** p<.001
Overall, workers stayed in Taiwan for more than 6 years have a higher score on affective commitment than all the other groups, see table 4.3. But if investigating the data from two national groups, there is no significance among Indonesia workers, while the Philippine workers worked less than one year score significantly higher than those worked more than four years (p=.028<.05), see table 4.6. The patterns are also shown in figure 4.1. This “higher
commitment in the very first year” phenomenon is similar to honeymoon phase of U-curve (Black and Mendenhall, 1991).
Since there is a relative little of time for a large set of negative reinforcements to have accumulated during the initial stage of working in Taiwan, migrant workers tends to have higher commitment.
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00
Year-in-Taiwan
Affective commitmen
Filipinos Indonesian All
Figure 4.1 Affective commitment patterns of migrant workers
Moreover, through Chi-square test of independence, it shows the salary and year-in-TW variables are not significantly related (χ2=19.942, p=.223). This means that these two variables are independent to each other; implicating that the salary of migrant workers does not change as their tenure becoming longer, which match with the practical pay system for migrant workers.
Except for the above-mentioned variables, this study also investigates the data from shift groups. The results show that night shift workers in A company (composed of Filipinos) have significantly lower scores on performance appraisal and rewards than dayshift workers.
However, in B company (composed of Indonesians), there is no significant difference between day and night shift workers.
Thus, hypothesis 1, hypothesized that there is a significant difference in affective commitment among different demographic groups of migrant workers in Taiwan, is partly supported. Except for marital status, education and religion, different nationality, gender, age, year-in-Taiwan groups have significant differences in affective commitment, see table 4.9.
Hypothesis 2, hypothesized that there is a significant difference in ECCC among different demographic groups of migrant workers in Taiwan, is rejected. None of the demographic groups of migrant workers have significant difference in ECCC, see table 4.9.
As for hypothesis 3, hypothesized that there is a significant difference in HR practices among different demographic groups of migrant workers in Taiwan, is partly supported.
Different marital status, age, year-in-Taiwan, education and religion groups of migrant workers have no significant difference in HR practices. However, different nationality, gender and salary groups have significant differences in HR practices, see table 4.9.
Correlation and Regression Analysis of Variables
Descriptive statistic and correlation of variables
Table 4.7 shows the descriptive statistic and the correlations of all variables. The correlations of four HR practices and affective commitment are between .264 and .421, which are significantly positively related (p<.01). The correlations of four HR practices and ECCC are between .206 and .432, which are significantly positively related (p<.01). The correlation of ECCC and affective commitment is .270, which is significantly positively related as well (p<.01). This means the better HR practices migrant workers feel, the higher their affective commitment to the organization. The better ECCC they have, the higher their affective commitment to the organization. Besides, the correlations of overall HR practices and ECCC, and affective commitment are .366 and .412 at 0.01 level.
Thus, hypothesis 4, hypothesized that there is a relationship between HR practices and extent of ECCC of migrant workers in Taiwan, is supported.
Since dependent variable is positively related to all independent variables (small to median-size correlation), and there is no correlations higher than 0.7 within independent variables, multiple regression is further conducted.
Table 4.7 Descriptive statistic and correlation of independent and dependent variables
Variable Mean SD 1 2 3 4 5 6
1. Training 3.77 .479 (0.899)
2. PA 3.42 .671 .638** (0.907)
3. Benefit 3.27 .578 .523** .498** (0.850)
4. Rewards 3.68 .616 .518** .599** .512** (0.795)
5. ECCC 3.52 .570 .335** .206** .432** .232** (0.859)
6. AC 3.50 .499 .421** .264** .321** .312** .270** (0.764) ( ) refers to Cronbach’s α of internal consistency
** p<.01, n=228,
Influence of HR practices and ECCC on affective commitment
Stepwise multiple regression is employed to test the influence of HR practices and ECCC on affective commitment. After entering all the independent variables (training, performance appraisal, benefit, reward and ECCC) and dependent variable (affective commitment) into model 1, the results in table 4.8 shows that only ECCC and one of the HR
practices – training could explain 19.6% of the variance (F=27.359, p<.001), and △R2 achieved the standards of significance (p<.05). Other HR practices – performance appraisal, benefit and reward) do not explain the variance significantly in this study. To see the standardized coefficients of each variable, it is found that both training and ECCC have a significantly positive beta value (.372 and .145), meaning that (1) the influence of training and ECCC are positive to affective commitment and (2) training has a bigger influence than ECCC on affective commitment. This implies that the better migrant workers’ perception of training in the organization, the higher they will commit to the organization affectively and the better their perception of ECCC, the higher their affective commitment as well.
Besides, tolerance, VIF (variance inflation factor) and CI (condition index) in model 1 are .888 (>.01), 1.127 (<10) and 19.407 (<30), which shows there is no problem of collinearity in model 1, see table 4.8.
Thus, hypothesis 5 is supported by the results. The better effective cross-cultural communication migrant workers have in Taiwan, the greater their affective commitment to the organization. As for hypothesis 6, three of HR practices – performance appraisal, benefit and reward are not supported in this study. Migrant workers’ perceptions of performance appraisal, benefit and reward do not influence their affective commitment significantly. Only training is significantly supported. The better training migrant workers in Taiwan perceive, the greater their affective commitment to the organization.
Table 4.8 Summary of regression analysis
Variables Model 1 Model 2
β value β value
Independent variable
Training .372*** .356***
Performance appraisal n.s. n.s.
Benefit n.s. n.s.
Moderate effect of ECCC
Model 2 in table 4.8 shows the results of testing hypothesis 7 – the extent of effective cross-cultural communication will moderate the relationship between HR practices and commitment significantly (F=21.21, p<.001) with △R2 achieving the standards of significance (p<.05). To see the standardized coefficients of interact item (training x ECCC), a negative beta is found (β=-.162, p<.05). This means the effect of interact item (training x ECCC) are negative to affective commitment. In other words, part of hypothesis 7 is supported. The extent of effective cross-cultural communication will moderate the relationship between training and affective commitment.
Besides, tolerance, VIF (variance inflation factor) and CI (condition index) in model 2 are .982 (>.01), 1.019 (<10) and 19.728 (<30), which shows there is no problem of collinearity in model 2, see table 4.8.
Figure 4.2 shows the moderate effect of ECCC to the relationship between training and affective commitment. It shows the slopes of fitting lines are both positive. This means that no matter migrant workers’ perception on ECCC is relatively high or low, the better they perceive training, the greater their affective commitment to the organization. But when ECCC is relatively low, the positive effect of training on affective commitment is stronger (β=.539);
while ECCC is relatively high, the positive effect of training on affective commitment is weaker (β=.245). This implies training has bigger positive influence on affective commitment while the migrant workers cross-cultural communication ability is relatively low. Thus, training can compensate the deficiency of communication. When migrant workers perceive relatively worse training, those with high ECCC have a high affective commitment than those with low ECCC. In contrast, when migrant workers perceive better training, those with low ECCC have a higher affective commitment than those with high ECCC, but the difference is much smaller. In figure 4.2, it shows migrant workers with better training and lower ECCC have the highest affective commitment.
Table 4.9 summarizes the results of hypotheses testing.
Figure 4.2 Moderate effect of ECCC
Low ECCC High ECCC
Low ECCC y = 0.539x + 1.485 High ECCC y = 0.245x + 2.619
1.5 2 2.5 3 3.5 4 4.5 5
1.5 2 2.5 3 3.5 4 4.5 5
Training
Low ECCC High ECCC High ECCC
Affective commitment
3
Training
Table 4.9 Summary of hypotheses testing results Hypotheses Nationality Gender
Marital
Status Age YIT Salary Edu. Religion H1: There is a significant difference in affective commitment
among different demographic groups of migrant workers in Taiwan.
Accepted Accepted Rejected Accepted Accepted Accepted Rejected Rejected
H2: There is a significant difference in extent of effective
cross-cultural communication among different demographic groups of migrant workers in Taiwan.
Rejected Rejected Rejected Rejected Rejected Rejected Rejected Rejected
H3: There is a significant difference in HR practices among different demographic groups of migrant workers in Taiwan.
partly Accepted
partly
Accepted Rejected Rejected Rejected partly
AcceptedRejected Rejected
H4: There is a relationship between HR practices and extent of effective cross-cultural
communication of migrant workers in Taiwan. accepted
H5: The better effective cross-cultural communication migrant workers in Taiwan, the
greater their affective commitment to the organization. accepted
H6: The better HR practices migrant workers in Taiwan perceive, the greater their
affective commitment to the organization. partly accepted
H7: Effective cross-cultural communication will moderate the relationship between
H7: Effective cross-cultural communication will moderate the relationship between