In this chapter, the SPSS 20 for Windows and SmartPLS were used to analyze data. The SPSS was applied for analyzing respondent information and acquiring descriptive data report.
The correlation among variables was also discussed by using the SPSS. The structural equation modeling (SEM) is a statistical technique to test and evaluate the casual relationships for entire model effects. Moreover, partial least squares SEM (PLS SEM) was applied in this study to test hypotheses. Compared with other statistical tool for testing SEM like AMOS or LISREL, PLS adopts different estimation methods, so the requirement of sample size is relatively small. Through using SPSS and SmartPLS, the hypotheses testing and descriptive data were showed to answer research questions.
Descriptive Statistics of e-HR Practices
Before answering the research questions, it is necessary to know the usage of e-HR practices in Taiwan. Since it was mentioned that e-HR practices are scattered and nonscientific (Lengnick-Hall & Moritz, 2003), the findings and discussion of e-HR practices were according to 204 companies to reveal existent e-HR practices. From the previous EFA result of e-HR practices, the e-HR practices were categorized as six dimensions that included:
training and development, performance management, human resource planning, recruiting and selection, employee benefits, and compensation and rewards.
In general, office software is the primary tool for companies to perform e-HR activities, especially in training and development (see Table 4.1), performance and management (see Table 4.2), and human resource planning (see Table 4.3) dimensions. In the dimension of recruiting and selection practices (see Table 4.4), about 37% of the companies apply packaged software to do external recruiting. However, prescreening and interviewing
54
applicants are conducted by manual operation, which indicates that these activities are not appropriate to be replaced by electronic processes. As for employee benefits (see Table 4.5), approximately 40% of companies allow employees to access their information through packaged and integrated software, while it is still not approved yet for employees to make changes of their own records. As for carrying out compensation and rewards activities (see Table 4.6), most sample companies use packaged software and integrated software, meaning that these work processes are more appropriate to be transformed to electronic processes. To be noted, Table 4.3 shows that about 40% of companies do not pay lots of attention to guide employees to develop their individual development plan, besides, results also shows about 37% of companies do not plan to conduct succession plan. The distribution of each dimension is summarized in Table 4.1 to Table 4.6.
Table 4.1.
Descriptive Statistics of e-HR Practices- Training and Development Item
No practice Manual Office software Packaged software Integrated software Mean S.D.
39. Assessing training needs 6.9% 28.9% 47.1% 8.8% 8.3% 1.96 0.88 40. Scheduling training
sessions
3.4% 26.5% 46.6% 14.7% 8.8% 2.06 0.89 41. Developing instructional
materials
5.9% 21.6% 53.4% 12.3% 6.9% 2.05 0.81
42. Delivering courseware to diverse locations
5.9% 20.1% 45.1% 19.1% 9.8% 2.2 0.89
43. Handling and notifying course application
1.5% 18.6% 40.2% 25.5% 14.2% 2.36 0.95
(continued)
55
Table 4.1. (continued) Item
No practice Manual Office software Packaged software Integrated software Mean S.D.
44. Recording employee attendance at each session
1.0% 21.1% 41.7% 21.6% 14.7% 2.3 0.97
45. Capturing the costs associated with each training session
3.4% 17.6% 43.1% 22.5% 13.2% 2.32 0.93
46. Evaluating training effectiveness
7.4% 21.1% 49.5% 16.2% 5.9% 2.07 0.81
47. Constructing the training blueprint according to different positions and needs
15.7% 21.6% 43.6% 13.2% 5.9% 2.04 0.83
48. Conducting the assessment of training need for
departments and organization
5.9% 22.7% 54.2% 11.3% 5.9% 2.01 0.79
49. Conducting personal training need analysis, such as skill inventory and the linkage behind the course
11.9% 20.8% 48.5% 11.4% 7.4% 2.06 0.84
56
Table 4.2.
Descriptive Statistics of e-HR Practices- Performance Management Items
No practice Manual Office software Packaged software Integrated software Mean S.D.
1.Conducting employee
entrance and exit surveys 23% 16.7% 25.5% 23% 11.8% 2.39 0.99 2. Creating access to an
employee handbook and posting related benefit policies and procedures
0% 16.7% 43.6% 22.1% 17.6% 2.41 .96
16. Producing performance documentation and employee evaluations
4.4% 15.3% 33% 29.6% 17.7% 2.52 .97
17. Notifying supervisors of the need to conduct scheduled employee performance reviews
3.9% 15.2% 37.7% 27% 16.2% 2.46 .95
18. Tracking progress toward goal attainment
9.3% 19.1% 33.8% 24.0% 13.7% 2.36 .98
20. Building and saving employee personal
performance improvement plan
14.2% 13.2% 36.8% 21.6% 14.2% 2.43 .94
28. Helping employees assess their performance and set goals
20.2% 14.8% 34.5% 20.2% 10.3% 2.33 .93 30. Summarizing information
from individual development plans
42.4% 12.8% 28.1% 13.8% 3.0% 2.12 .81
57
Table 4.3.
Descriptive Statistics of e-HR Practices- Human Resource Planning Item
No practice Manual Office software Packaged software Integrated software Mean S.D.
12. Conducting job evaluations 21.1% 23% 34.3% 14.2% 7.4% 2.07 0.92 22. Generating organizational
charts 3.4% 15.2% 63.2% 12.3 5.9% 2.09 .72
23. Building schedules based on both staffing requirements and employee requests
9.3% 20.1% 55.9% 11.3% 3.4% 1.98 0.71 24. Tracking labor costs 9.4% 14.3% 52.2% 12.8% 11.3% 2.23 0.87 25. Tracking and optimizing
the distribution of human resources and their productivity
19.7% 17.2% 47.8% 8.4% 6.9% 2.06 .081
26. Conducting succession planning
34.8% 23.5% 32.8% 5.4% 3.4% 1.83 0.79 27. Providing employees with
self-assessment and career development guidance
40.2% 16.7% 26.0% 12.7% 4.4% 2.08 0.89
58
Table 4.4.
Descriptive Statistics of e-HR Practices- Recruiting and Selection Item
No practice Manual Office software Packaged software Integrated software Mean S.D.
31. Submitting job requisition
information 2.0% 33.3% 41.7% 13.2% 9.8% 2.00 0.94
33. Recruiting outside the organization
2.9% 17.2% 28.9% 37.3% 13.7% 2.49 0.94 34. Tracking applicants
through the staffing process
7.4% 24.5% 32.8% 26.5% 8.8% 2.21 0.94 35. Processing resumes 1.0% 27.0% 34.8% 29.4% 7.8% 2.18 0.93 36. Prescreening job applicants 1.0% 32.8% 29.4% 28.9% 7.8% 2.12 0.97
Table 4.5.
Descriptive Statistics of e-HR Practices- Employee Benefits Item
No practice Manual Office software Packaged software Integrated software Mean S.D.
6. Letting employees make changes to their own benefits records
41.2% 16.2% 9.3% 18.1% 15.2% 2.55 1.15
7. Processing of monthly benefits invoices
7.4% 36.3% 16.7% 19.6% 20.1% 2.25 1.19 10. Allowing employees to
access pay data information
34.8% 11.8% 6.4% 23.5% 23.5% 2.9 1.09
59
Table 4.6.
Descriptive Statistics of e-HR Practices- Compensation and Rewards Item
No practice Manual Office software Packaged software Integrated software Mean S.D.
9. Transferring employee data between HR and outside payroll systems
1.0% 2.5% 5.9% 47.5% 42.6% 3.33 0.7
11. Performing compensation structure and cost analysis
9.8% 10.3% 31.9% 24.5% 23.5% 2.68 0.99 14. Processing employee
compensation relevant data such as ages, vacation and sick time
1.0% 1.5% 7.4% 46.6% 43.6% 3.34 0.68
19. Collecting employee time and attendance data
0.5% 4.4% 7.4% 48.0% 39.7% 3.24 0.77
60
Correlations
To estimate the initial relationship among all the research variables, the Pearson’s correlation coefficient was analyzed to acquire general results. In the following Table 4.7, means, standard deviations, and correlations among variables are shown. The Cronbach’s alphas are also reported on the diagonal.
According to the report, competitive tension has a positive correlation with e-HR practices, however, the correlation coefficient is not significant. In addition, the correlations between competitive tension and two financial indicators, ROA (r=-0.184, p<0.01), and EPS (r=-0.209, p<0.01), are negative. As expected, strategic leadership (r=0.242, p<0.01) and IT capability (r=0.210, p<0.01) have positive and significant correlations with e-HR practice.
Furthermore, the e-HR practice has a negative and significant relationship with HR efficiency (r=-0.246, p<0.01) meaning that when a company uses more e-HR practices, the employee to human resource personnel ratio is lower. This is opposite to the original research assumption. As for the financial performance, the result indicates that e-HR practice is positively related to ROA (r=0.179, p<0.05) but not related to EPS, meaning the more usage of e-HR practices, the better performance of the company’s ROA.
61
Table 4.7.
Correlation Analysis Results
Variables Mean S. D. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1.Org. size (Log_capital) .58 9.96
2.Industry sector .73 .45 .043
3.Competitive tension 47.61 23.31 .026 .306**
4.Strategic leadership 5.55 .997 -.054 -.103 -.149* (0.97)
5.IT capability 4.83 .89 -.035 -.130 -.067 .353** (0.92)
6.EHR_All 2.25 .52 .031 -.122 .069 .242** .210** (0.98)
7.Training & development 2.11 .69 .087 -.068 .017 .268** .169* .752** (0.94)
8.Performance mgt. 2.35 .75 .035 -.168* .056 .215** .210** .789** .418** (0.93)
9.Human resource planning 2.05 .60 -.051 -.104 -.080 .168* .194** .757** .492** .583** (0.89) 10.Recruiting & selection 2.19 .78 -.006 .006 .146* .212** .092 .675** .408** .416** .462** (0.88) 11.Employee benefit 2.44 1.05 -.083 -.174* .051 .085 .022 .596** .291** .499** .340** .279** (0.78) 12.Compensation & reward 3.15 .61 .092 .020 .023 .105 .216** .533** .266** .405** .353** .298** .402** (0.77) 13.HR efficiency 4.29 2.01 .024 .147* .062 -.229** -.065 -.246** -.215** -.277** -.186* -.129 -.094 -.038
14.ROA 6.40 10.81 -.039 -.014 -.184** .074 .039 .179* .186** .072 .123 .192** .130 .031 -.013
15.EPS 1.58 3.76 -.064 -.198** -.209** .016 .027 .135 .138 .125 .120 .064 .090 .037 .002 .759**
Note. 1. *p<0.05, **P<0.01, ***p<0.001
2. Strategic leadership and IT capability adopted 7-point Likert scale.
3. Coding for industry sector: 0 for non-manufactory industry; 1 for manufactory industry.
4. Coding for variable 6~12: 1 for manual; 2 for office software; 3 for package software; 4 for integrated software.
62
PLS Model Testing
The partial least square structural equation modeling (PLS-SEM) technique was applied in statistical analysis using SmartPLS software to test research hypotheses. According to the research of Hair, Ringle, and Sarstedt in 2011, if the structural model is complicated, meaning with many constructs or many indicators, using PLS-SEM is more appropriate.
The dimensions were used as indicators of main study variables when analyzing research hypotheses to maintain parsimony of the PLS model, since there were many items in some of the variables. Hence, the strategic leadership adopted two dimensions based on its original scale and IT capability also adopted two dimensions. To be noted, six dimensions were used to represent e-HR practices. Two control variables, organization size and industry sector, were included to precisely examine the research model.
As the validity and reliability analyses of measurement in previous chapter has ensured valid and reliable measurement models, thus, the structural model can be further analyzed by using SmartPLS. The result of hypotheses testing is explained by path coefficients (β) and R square. The path coefficients is used to decide the relationship or degree of influence between outcome variables and predictor variables, while the R square indicates how much of the variance of dependent variable can be explained by the independent variables. The significance of path coefficient is evaluated by the t-value, which is retrieved after running the bootstrapping procedure. During the bootstrapping procedure, the original samples are duplicated and randomly selected to form a new sample for testing hypothesis more precisely. The minimum number of bootstrapping samples is 5000 (Hair et al., 2011). The rules of thumb for interpreting structural model testing results are summarized in Table 4.8.
63
Table 4.8.
The Rules of Thumb for Structural Modeling Testing
Testing Standard
1.R square (R2) (A) R2 > .19 (weak); R2 > .33 (moderate); R2 > .67 (strong) (Chin, 1998)
(B) R2 > .02 (weak); R2 > .13 (moderate); R2 > .26 (strong) (Cohen, 1988)
2.Path coefficients (β) Minimum number of bootstrap sample: 5000
After Bootstrapping, t-value > 1.65 (weak); t-value > 1.96 (moderate); t-value > 2.58 (strong)
(Hair et al., 2011) 3.Goodness of Fit index
(Gof index)
0.1= small, 0.25= medium, and 0.36= large (Wetzels et al., 2009)
After SmartPLS structural model testing, the relationship among predictor variables, outcome variables, and control variables are organized in Table 4.9. This table shows the path coefficient, Error, and t-value of a relationship to examine the predictive power of each independent variable, and summarizes the R square for each outcome variable to show how much variance of an outcome variable was explained in the study model.
The effects from antecedents to e-HR adoption explain 9.9% of model variance (R2
=0.099). According to Chin (1998), the R2 is lower than 0.19, meaning the predictive power is too low to explain the model; however, based on the rule of Cohen (1988), the R2 exceeds 0.02, which indicates weak predictive power. The effect from e-HR adoption to HR efficiency explains 21.7% of model variance (R2 =0.217), which means weak predictive power according to Chin (1998), and moderate predictive power according to Cohen (1988).
As for the effects from e-HR adoption to ROA and EPS, each explains 2.4% and 12% of
64
model variance respectively (R2 =0.024 and 0.120), both of which represent weak predictive power according to Cohen (1988). The result of model testing is shown in Figure 4.1.
Table 4.9.
Path Coefficients, Error, t-values and R square Original
Financial performance-EPS 0.013 0.013 0.012 0.012 1.098 Competitive tension→
Financial performance-ROA 0.010 0.010 0.011 0.011 0.912 Competitive tension→ HR efficiency -0.018 -0.019 0.016 0.016 1.133 Competitive tension→ e-HR practice 0.097 0.099 0.073 0.073 1.331 Control_industry
Control_industry→
Financial performance-EPS -0.287 -0.286 0.056 0.056 5.135***
Control_industry→
Financial performance-ROA -0.079 -0.081 0.071 0.071 1.108 Control_industry→ HR efficiency 0.068 0.069 0.058 0.058 1.189 Control_org. size
Control_org. size→
Financial performance-EPS -0.212 -0.212 0.054 0.054 3.943***
Control_org. size→
Financial performance-ROA -0.119 -0.119 0.068 0.068 1.743*
Control_org. size→ HR efficiency -0.372 -0.372 0.058 0.058 6.402***
IT capability IT capability→
Financial performance-EPS 0.024 0.026 0.016 0.016 1.485 IT capability→
Financial performance-ROA 0.018 0.019 0.016 0.016 1.105 IT capability→ HR efficiency -0.033 -0.035 0.018 0.018 1.825*
IT capability→ e-HR practice 0.181 0.192 0.081 0.081 2.240**
(continued)
65
Table 4.9. (continued)
Original Sample
(O)
Sample Mean
(M)
Standard Deviation (STDEV)
Standard Error (STERR)
T Statistics (|O/STERR|)
R square Strategic leadership
Strategic leadership→
Financial performance-EPS 0.028 0.030 0.016 0.016 1.817*
Strategic leadership→
Financial performance-ROA 0.021 0.022 0.017 0.017 1.232 Strategic leadership→ HR efficiency -0.038 -0.041 0.020 0.020 1.951*
Strategic leadership→ e-HR practice 0.211 0.217 0.060 0.060 3.498***
e-HR practice e-HR practice→
Financial performance-EPS 0.135 0.136 0.060 0.060 2.240**
e-HR practice→
Financial performance-ROA 0.099 0.100 0.070 0.070 1.410 e-HR practice→ HR efficiency -0.181 -0.187 0.062 0.062 2.907***
Outcome variables
e-HR practice 0.099
HR efficiency 0.217
Financial performance-ROA 0.024
Financial performance-EPS 0.120
66
Note. 1. ß stands for the path coefficients
2. *t-value>1.65, **t-value>1.96, ***t-value>2.58 Figure 4.1. Hypotheses testing with path coefficient and R square
According to Figure 4.1, the first hypothesis asserts that competitive tension of the company in the industry has an influence on the practices of e-HR, and this assumption is not supported (β=0.097; t<1.65). The second hypothesis posits that strategic leadership of the company has a positive influence on the practices of e-HR, and it is supported (β=0.210;
t>2.58). The third one is that IT capability of the company has a positive influence on the practices of e-HR, it is also supported (β= 0.181; t>1.96). Hypothesis four is investigated by hypothesis 4a, 4b, and 4c. The hypothesis 4a indicates practices of e-HR have positive influence on HR efficiency. However, this hypothesis is not supported (β= -0.181; t>2.58).
The path coefficient is negative, meaning the usage of e-HR practices have negative influence on HR efficiency, which was measured by the employee-HR personnel ratio. The hypothesis 4b indicates practices of e-HR have an impact on the ROA of the company, and it is not supported (β= 0.099; t<1.65). The last hypothesis 4c indicates practices of e-HR have a
67
positive influence on the EPS of the company; this hypothesis is supported (β= 0.135; t>1.96).
The overview of hypothesis testing is organized in Table 4.10.
Table 4.10.
The Overview of Hypotheses Testing
Hypothesis Path
coefficient T-value Result Hypothesis 1: Competitive tension of the firm in
the industry has a positive influence on the practices of e-HR.
0.097 1.323
Not supported Hypothesis 2: Strategic leadership of the firm
has a positive influence on the practices of e-HR 0.210 3.605*** Supported Hypothesis 3: IT capability of the firm has a
positive influence on the practices of e-HR. 0.181 2.278** Supported Hypothesis 4: Practices of e-HR has a positive
influence on the organization outcomes.
Hypothesis 4a: Practices of e-HR has a
positive influence on the HR efficiency. -0.181 2.831***
Not supported Hypothesis 4b: Practices of e-HR has a
positive influence on the ROA of the firm. 0.099 1.382
Not supported Hypothesis 4c: Practices of e-HR has a
positive influence on the EPS of the firm. 0.135 2.205** Supported Note. *t-value>1.65 (weak), **t-value>1.96 (moderate), ***t-value>2.58 (strong)
In general, the antecedents of e-HR practices are supported except competitive tension.
The empirical data support the literature. However, only EPS as the organization outcome is
68
supported. In the hypothesis of HR efficiency as an outcome, the data show opposite evidence to what was originally proposed.
As for the goodness of fit (Gof) index, Tenenhaus, Amato, and Esposito (2004) developed this index and took both measurement model and structural model into consideration for measuring the model performance. They also posited that the square root of average AVE multiplied by average R2 is the Gof index. The rules of thumb for Gof index is:
0.1 equal to small, 0.25 equal to medium, and 0.36 equal to large; and those number present different degree of how data fit the research model.
Thus, the Gof index in this model is calculating as follows: square root of average AVE of all constructs (0.9295) multiplied by average R2 of all endogenous constructs (0.34) equals to the Gof index 0.31 which means the data have medium fit (Gof index > 0.25) to this model.
69