In this chapter, the findings of the study are revealed with the overall highlights of the analysis processes in the confirmatory factor analysis, reliability, correlation, simple linear regression analysis as well as hierarchical regression analysis utilized and the demographics of the multiple populations studied. In a later part, the research results and the discussions are introduced.
Demographic Statistics
Sample Characteristics for HR Managers
On account of the fact that the pair design was applied in this research, the data were drawn from multiple respondents in the targeted study population in Mongolia. Overall, putting a tremendous effort from the researcher brought data from 90 different private sectors in Mongolia including the dataset of 90 HR managers as well as 270 employees in this study.
In the demographic statistics for HR managers, the sample characteristics of gender, career seniority, job title, and education level were contained to release the background information about the representative sample size of HR managers in Mongolia. Table 4.1 provides the summary of the profile of HR professionals.
Among the 90 HR participants, 21 males (23.3%) and 69 females (76.7%) participated in this study. Altogether 36 participants (40 %) of the total respondents have 3-5 years of career seniority in an HR related position in their current company. With regards to the rest of respondents, 33 have 1-2 years (36.7%), 16 have 6-10 years (17.8%), 2 have 11-15 years (2.2%), 2 have 16-20 years (2.2%), while only 1 participant responded that she has 21 years and above HR related job seniority in her current organization (1.1%). Also, the data drew that the most of HR survey respondents were either HR staffs (44.4%) or HR managers (32.2%).
Apart from the majority, there were 7 participants at HR supervisor positions (7.8%) and 14 at HR professional (14 %). Of the respondents 35 have not possessed a degree or vocational training in HR (38.9%), 17 have completed only a vocational course or training program in HR (18.9%), 26 have a bachelor degree (28.9%), and 12 have a Master degree (13.3%) in HR.
47
Table 4.1.
Demographics for HR Managers
Variable Category Frequency %
HR position
HR chief 7 7.8
HR professional 14 15.6
HR manager 29 32.2
HR staff 40 44.4
Career_Seniority
1-2 years 33 36.7
3-5 years 36 40.0
6-10 years 16 17.8
11-15 years 2 2.2
16-20 years 2 2.2
21 or above 1 1.1
Gender Male 21 23.3
Female 69 76.7
Education level
Non-Vocational course/degree in HR 35 38.9 Only vocational course/Training in HR 17 18.9
Bachelor degree in HR 26 28.9
Master degree in HR 12 13.3
Note. N(HR managers)=90.
With regard to the demographic statistics for employees, a few questions about the personal characteristics (gender, job level, and career seniority as well as education level) were asked to reveal the background details about the representative sample size of employees in Mongolia.
In Table 4.2 an overview of the profile of employee is shown.
Among the 270 employees who participated in this study, a total of 166 participants were female (64.5%) and 104 were male participants (38.5%). The most of the data was gathered from 186 respondents at the staff positions (68.9%) compared 84 respondents at supervisor positions (31.1%). Besides, a total of 127 respondents reported with having 1-2 years of service at their current organization (47%), 73 reported 3-5 years of career seniority (27%), 43 reported in 6-10 years of career seniority (15.9%) while altogether 10 reported as having 16-20 as well as 21 or above years of career seniority in their current organizations (3% and 0.7%
respectively). Along with this, a total of 191 participants earned a bachelor degree (70.7%), 32
48
earned a Master degree (11.9%), and only 1 earned a Ph.D. degree (0.4%) while altogether 46 respondents earned junior high school or below, senior high school and college or specialized training program (23.3%).
Table 4.2.
Demographics for Employees
Variable Category Frequency %
Position Supervisor 84 31.1
Staff 186 68.9
Career_Seniority 1-2 years 127 47.0
3-5 years 73 27.0
6-10 years 43 15.9
11-15 years 17 6.3
16-20 years 8 3.0
21 or above 2 0.7
Gender Male 104 38.5
Female 166 61.5
Education level Junior High School or Below 8 3.0
Senior High School 2 0.7
College or Specialized training
program 36 13.3
Bachelor degree 191 70.7
Master degree 32 11.9
PhD 1 0.4
Note. N(employee)=270.
49
Validity and Reliability
To address face and content validity, all measurements utilized in this study are previously established scale in the existing literature. Owing to the fact that all established scales were initially established and published in English, the back translation, the peer-review as well as the expert review were conducted in this research. The confirmatory factor analysis (CFA) is another type of structural equation modeling (SEM) and an essential tool of construct validity, usually attempts to deal with the measurement models (Brown, & Moore, 2012). One of the common goals of these two approaches is to assess whether a specified model represents or fits the data collected in the study (Hooper, Coughlan, & Mullen, 2008). Following this, CFA also determines "quality of the factor structure by statistically testing the significance of the overall model and items loadings on factors" (Hinkin, 1998, p. 13).
Confirmatory Factor Analysis
The Confirmatory Factor Analysis (CFA) was conducted to check the validity of the measurement model for each variable of this study in Mplus7. When items on the measurement model have a factor loading lower than 0.4, were considered to be the poor items to measure the construct validity and were deleted from the measurement model (Hinkin, 1998; Kim &
Mueller, 1978). The items were removed based on the following criteria and conditions in this study: first, the model fitness indices were examined for each variable whether these fit-indices were achieved the required levels. The criteria for the good model–fit indices are shown in Table 4.3. Particularly, once the model fit-indices are met the satisfactory levels, these items which have a lower factor loading than 0.4, are still kept in the measurement model such as second items of HR Role (R2). Secondly, these items HS1, S5, S10, S11, C1, C3 and C5 that produced non-significant value at p>.05, were deleted in the first place. Finally, the statement of each item with low factor loading was reviewed by the researcher based on the importance of the item along with the purpose of this study, for instance, the second item of Challenge stressors (CS2) has remained in the measurement model.
50
Table 4.3.
Summary of Good-Fit Criteria
Fit index Acceptable levels
Chi-Square X2 Less than. 05 (Hooper et al., 2008) CFI and TLI > .90 Reasonable fit
> .95 Good model fit
(Hu &Bentler, 1999)
RMSEA < .03 Excellent fit
< .05 Good model fit
> 05 but <. 08 Reasonable fit
(MacCallum, Browne, &
Sugawara, 1996; Hu & Bentler, 1999)
SRMR 0 indicates an Excellent fit
>. 05 Good fit
> .05 but <. 08 Reasonable fit
(Byrne, 1998; Diamantopoulos and Siguaw, 2000)
Note. Adapted from “Structural Equation Modelling: Guidelines for Determining Model Fit.”
By D. Hooper, J. Coughlan, and M. Mullen, 2008. Economic Journal of Business Research Methods, 6(1), p. 53-60. Copyright 2008 by the Academic Conferences Ltd.
On account of following two reasons, each variable was selected to run separately in Mplus in order to produce CFA results accurately; first, since the pair design was utilized in this study, data from multiple respondents were entered into two datasets in SPSS, which restricted the researcher to have all variables performed together in Mplus at the same time. Secondly, the number of total items used in this study was 113, which exceeded the capacity of Mplus to be functioned accurately. To address the limitations mentioned above, the item parcel was applied, and each variable was employed and reported separately in Table 4.5-Table 4.7.
The first round of the CFA, all items of each variable were entered in Mplus and estimated.
However, the results did not pass the requirements for model-fit indexes, except HR Competencies. Therefore, several items were deleted, and modifications were applied to the original scales of Hindrance stressors and HR Effectiveness based on the value of the factor loading and significance of each item in the model. After the modification stage in each model, the second round of the CFA was conducted and revealed the acceptable model fit indexes, illustrated in Table 4.4. Consequently, all models were significant, and all these indexes passed the fundamental criteria for the acceptable fit for each measurement model.
51
Table 4.4.
Model-Fit Indices for Alternative Factor Models of HR Competencies, Challenge Stressors, Hindrance Stressors and HR Effectiveness
Variables X2 Df CFI TLI RMSEA SRMR
HR Competencies 214.9* 137 .927 .909 .085 .049
HR Effectiveness 235.6* 149 .954 .947 .080 .052
Challenge and Hindrance stressors
15.1 13 .982 .971 .043 .050
Note. X2=chi square goodness of fit statistic. df=degree of freedom. RMSEA= Root-Mean-Square Error of Approximation. AIC=Akaike Information Criterion. CFI= Comparative Fit Index. TLI=Tucker Lewis Index. SRMR=Standardized Square Root Mean Residual. *p.05.
HR competencies. In the first round of the CFA, a total of 19 parceled items of HR Competencies went through the confirmatory factor analysis in Mplus to check the validity of the measurement. The result presented that the model fit indices achieved the required levels, as shown in Table 4.5. In addition to the model fit results, all factor loadings were statistically significant, and the standardized loading estimations were all above 0.4 (ranging from 0.67 to 0.88). Moreover, all the constructs’ AVEs for sub-dimensions of HR competencies were greater than 0.5 (ranging from 0.47 to 0.77), except HRCCA (Human Resource Competency Credible Activist) at 0.47, however, which is very close to 0.5 considered to be accepted in the model. Altogether, it can be assured that the convergent validity was established since AVE is greater than 0.5 and CR is larger than 0.6 (Hair, Black, Babin & Anderson, 2010). The overall fit of this measurement model was X2 (137) = 214.9; X2/ d.f=1.56; CFI and TLP were 0.92 and 0.90 respectively. Besides, SRMR and RMSEA were 0.04 and 0.08 respectively. All things considered, all these indexes of HR Competencies indicated an acceptable fit for the measurement model.
52
Table 4.5.
CFA Results for HR Competencies
Variables Means Factor
Loadings
Note. N(HR managers)=90. Factor loading > 0.4, Average Variance Extracted (AVE)>0.5, and Composite Reliability (CR)>0.6 are acceptable. All factor loadings are significant at p.001.
The parceled items were applied.
53
HR effectiveness. In the first round of CFA test, 30 items of HR Effectiveness went through the confirmatory factor analysis in Mplus to check the validity of the measurement. The result presented that the model fit indices failed to achieve the required levels, and 11 items on three dimensions of HR Effectiveness were modified concerning the outputs of the modification indexes. Specifically, 6 items from HR service; S1(0.273), S5(0.126, not significant), S8 (0.385), S9 (0.395), S10 (-0.016, not significant), S11(-0.029, not significant), and 5 items from the HR contribution; C1 (0.061, not significant), C3 (-0.039, not significant), C4 (0.182), C5 (0.109, not significant), C10 (0.22) were deleted in order to improve the model-fit indexes for the measurement model. Also, all items of HR Role were kept even though the second item (R2) was not qualified to be retained in the model due to the poor factor loading at 0.309. After the modification stage, AVE and CR were produced at the satisfactory levels of 0.7 and 0.9 respectively, as illustrated in Table 4.6. The overall fit of this measurement model was X2 (149)
=235.6; X2/ d.f =1.58; CFI and TLP were 0.95 and 0.94 respectively. Also, SRMR and RMSEA were 0.05 and 0.08 respectively. Taking everything into consideration, all these indexes of HR Effectiveness indicated an acceptable fit for the measurement model.
54
Table 4.6.
Modified CFA Result for HR Effectiveness
№ Variables Means Factor
Loadings
Reliability Coefficient
AVE
Service (S) 0.95 0.723
1 S2 4.504 0.784
2 S3 4.719 0.828
3 S4 4.700 0.871
4 S6 4.537 0.869
5 S7 4.490 0.810
6 S12 4.726 0.899
7 S13 4.644 0.902
8 S14 4.526 0.878
9 S15 4.604 0.805
Role (R ) 0.915 0.704
10 R1 6.285 0.870
11 R2 4.926 0.307
12 R3 6.259 0.936
13 R4 6.211 0.943
14 R5 6.485 0.951
Contribution (C) 0.955 0.812
15 C2 4.363 0.765
16 C6 4.552 0.929
17 C7 4.556 0.954
18 C8 4.504 0.860
19 C9 4.563 0.982
Note. N(employees)=270. Factor loading >0.4, the assessment for Convergent Validity (AVE)>0.5, and Composite Reliability (CR)>0.6 are acceptable. All factor loading are significant at p<.001 except R2 at p<.01.
Challenge and Hindrance Stressors. In the first round of CFA test, eight items of Challenge and Hindrance stressors went through the confirmatory factor analysis in Mplus to check the validity of the measurement. The result presented that the model fit indices failed to achieve the required criteria. Thus, the first item of Hindrance stressors (HS1=0.069, not
55
significant) was removed from the measurement model in order to improve the model-fit indexes. Even though the second item (CS2) of Challenge stressors was not qualified to be retained because of the poor factor loading at 0.379, it is considered to be kept in the measurement model owing to the importance of the statement with the purpose of this study.
After the modification stage, AVE and CR were produced at the acceptable levels of 0.42 (closer to 0.5) and 0.78 respectively, illustrated in Table 4.7. The overall fit of this measurement model was X2 (13) =15.1; X2/ d.f=1.16; CFI and TLP were 0.98 and 0.97 respectively. In addition, SRMR and RMSEA were 0.05 and 0.04 respectively. Overall, all these good model fit indices of Challenge and Hindrance stressors indicated an acceptable fit for the measurement model.
Table 4.7.
Modified CFA Results for Challenge and Hindrance Stressors
Variables Means Factor
Loadings
Reliability Coefficient
AVE
0.78 0.42
Challenge stressors (CS) 0.78 0.34
CS1 3.88 0.54
CS2 3.04 0.37
CS3 3.55 0.86
CS4 3.67 0.44
Hindrance stressors (HS) 0.75 0.52
HS2 2.88 0.47
HS3 2.57 0.73
HS4 2.88 0.91
Note. N(HR Managers)=90. Factor loading >0.4, the assessment for Convergent Validity (AVE)>0.5, and Composite Reliability (CR)>0.6 are acceptable. All factor loading are significant at p<.001.
56
Discriminant validity. Discriminant validity is the extent to which a construct is truely divergent from all other constructs in the study model. Basically, it is tested by comparing AVE for any other two contructs with the square of the correlation scores between them (Hair et al., 2010). More specifically, if all variables’ AVE exceed the squarred correlation scores of the pairs of variables; hence, the discriminant validity is established. According to Table 4.8, all variables’ AVE in the model of this study were found to be larger than the squared correlation scores of the pairs of variables. Therefore, it is possible to conclude that the discriminant validity of the study model was confirmed to be established.
Table 4.8.
The Result of the Discriminant Validity
Variable AVE AVE is larger than the squared correlation scores of the pairs of variables
Overall HR Competencies 0.61 >
HRCSP 0.77 >
HRCCA 0.47 >
HRCCB 0.67 >
HRCCC 0.56 >
HRCII 0.71 >
HRCTP 0.65 >
Challenge stressors 0.34 >
Hindrance stressors 0.52 >
HR service 0.72 >
HR Role 0.70 >
HR contribution 0.81 >
Note. < = AVE is smaller than all squared correlation scores of the pairs of variables. >= AVE is larger than all squared correlation scores of the pairs of variables. N(HR managers)=90, N(employees)=270.
57
Reliability Analysis
According to the results of the reliability analysis, reported in Table 4.9, The Cronbach alpha’s values of the six dimensions of HR Competencies and three dimensions of HR Effectiveness were higher than 0.70 (ranged from 0.84 to 0.90), which is considered to be modest internal consistency. However, the internal consistency with a Cronbach alpha of 0.6-0.7 is also considered to be acceptable (Churchill, 1979; George & Mallery, 2003). Therefore, the Cronbach alpha of Challenge stressors at 0.61 was accepted to be qualified for the internal consistency test as a variable in this study.
Table 4.9.
Cronbach’s Alpha Analysis
Variables Number of items Cronbach's Alpha
Overall HR Competencies 75 .97
Strategic Positioner 12 .87
Credible Activist 13 .84
Capability Builder 11 .90
Change Champion 8 .89
Innovator and Integrator 20 .95
Technology Proponent 11 .89
Overall Challenge Stressors 4 .61
Overall Hindrance Stressors 3 .74
Overall HR Effectiveness 19 .96
Service 9 .93
Role 5 .92
Contribution 5 .90
Note. N(HR managers)=90, N(employee)=270.
Common Method Variance
Common Method Variance (CMV) is the way to observe and attempt to determine the biases which may proceed from the different circumstances, such as the survey instrument design, complexity, ambiguity, scale, format, the item’s context, a rater’s motivation and the length of survey instruments (Eichhorn, 2014).
58
In order to check common method variance (CMV), Harman's single factor test was utilized.
The results (Table 4.10) showed that the percentage of variance of each factor is smaller than 50%, which indicated that no a single factor accounts for the majority of the variance in the variables.
Table 4.10.
The Results of Harman’s Single Factor Score
Initial Eigenvalues Extraction Sums of Squared Loadings
Item Total
% of Variance
Cumulative
% Total
% of Variance
Cumula tive % HR Competencies 75 26.240 34.987 34.987 26.240 34.987 34.987
Challenge and Hindrance
stressors
7 2.314 33.064 33.064 2.314 33.064 33.064
HR effectiveness 19 19.769 34.682 34.682 19.769 34.682 34.682 Note. Extraction Method: Principal Component Analysis.
Intraclass Correlation Coefficient Analysis
An intraclass correlation coefficient (ICC) can be a common calculation of inter-rater reliability on quantitative data. In other words, it is a common statistical method to estimate the degree of agreement among various raters. Since the pair design was employed in this study, the ICC was conducted to assess the consistency of measurements rated by three emplyees measuring the same .
The result of the ICC analysis reported in Table 4.11 provided that a high degree of reliability was found among raters in this study. The average measure ICC was .964 with a 95%
confidence interval from .953 to .974 (F(88,4928)= 28.141, p<.001). Particularly, if the ICC is above 0.74, it is considered to be good reliability in the social science (Portney, & Watkins, 2000). More specifically, the result of ICC analysis in this study confirmed that three raters (employees) who have similar perception of the effectiveness of their HR professional and therefore we aggregate those individual evaluations for later analysis .
59
Table 4.11.
Intraclass Correlation Coefficient Analysis
Intraclass Correlationb
95 %
Confidence interval
F Test with True Value 0
Lower Bound
Upper Bound
Value df1 df2 sig
Single Measures .323a .263 .399 28.141 88 4928 0.000
Average Measures .964c .953 .974 28.141 88 4928 0.000 Note. Two-way mixed model where people effects are random and measures effects are fixed, N(employees)=270.
a. The estimator is the same, whether the interaction effect is present or not
b. Type C interclass correlation coefficients using a consistency definition. The between-measure variance is excluded from
c. This estimate is computed assuming the interaction effect is absent, because it is not estimable otherwise.
Correlation Analysis
The Pearson coefficient correlation analysis was performed to examine the correlations among sub-dimensions of independent and dependent variables as well as moderators in the theoretical framework of this study.
According to the results of the correlation analysis, shown in Table 4.12, Strategic Positioner, Credible Activist, Capability builder, Change Champion, Innovator/Integrator, and Technology Proponent have a significant positive relationship with overall HR Effectiveness, and the correlation values of them were between .005 and .038, which clearly indicates that the six dimensions of HR Competencies are highly correlated with overall HR Effectiveness. As for an overall correlation between independent and dependent variable, HR Competencies is positively correlated with HR Effectiveness, (r= 0.302, p<.01); however, HR Service and Role these are two sub-dimensions of HR Effectiveness, are not significantly correlated with two sub-dimensions of HR Competencies named HR Innovator and Technology Proponent (r=.203, n.s. for HR service and HR innovator, r=.149, n.s. for HR service and Technology proponent, r=.193, n.s. for HR role and HR innovator, r=.186, n.s. for HR role and Technology proponent, n.s). Similarly, HR Service is also not significantly correlated with a sub-dimension of HR Effectiveness named credible Activist (r=.147, n.s.). In contrast to HR Service and HR Role,
60
HR Contribution has a positive relationship with each sub-dimension of HR Competencies with correlation values between .001 and .015.
As for the moderating variables, the challenge stressors is positively correlated with only one sub-dimension of HR Competencies named Technology Proponent, (r=0.308, p<.01).
Except for the above-reported correlation, Challenge, as well as Hindrance stressors, do not have a significant and positive relationship with the dependent variable.
61
Table 4.12.
Pearson Correlation Analysis
Mean S.D. 1 2 3 4 5 6 7 8 9 10 11 12 13
1 Strategic Positioner 3.58 0.542
2 Credible Activist 3.91 0.423 .599**
3 Capability Builder 3.71 0.568 .729** .745**
4 Change Champion 3.71 0.521 .413** .621** .691**
5 HR Innovator 3.6 0.62 .632** .689** .757** .673**
6
Technology
Proponent 3.56 0.645 .678** .603** .690** .483** .737**
7
Overall HR
Competencies 3.68 0.467 .805** .825** .912** .760** .896** .846**
8 Challenge Sressors 3.54 0.572 .028 .114 .102 .119 .146 .308** .169
9 Hindrance Stressors 2.78 0.904 .026 .096 .077 .062 .149 .116 .106 .180
10 Overall_Stressors 3.16 0.576 .034 .132 .111 .108 .189 .243* .167 .637** .873**
11 Service 4.6 0.892 .243* .147 .250* .245* .203 .149 .245* .126 .158 .186
12 Role 6.25 1.53 .257* .209* .250* .245* .193 .186 .263* .109 .187 .201 .782**
13 Contribution 4.52 0.868 .329** .255* .312** .280** .275** .304** .349** .118 .120 .152 .786** .801**
14
Overall HR
Effectiveness 5.13 1.02 .293** .220* .286** .274** .233* .222* .302** .124 .174 .198 .905** .955** .913**
Note. **. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
62
Hypothesis Tests
Simple Linear Regression Analysis
Apart from the correlation analysis, Hypothesis 1 and its nine sub-hypotheses from H1:1 to H1:9 were tested by conducting the simple linear regression. Based on the proposed relationship direction between sub-dimensions of independent and dependent variables, the results of these nine hypotheses were split into three tables shown in Table 4.13- Table 4.15.
H1: HR competencies are positively related to HR effectiveness. A simple linear regression analysis was carried out to predict the overall HR Effectiveness based on overall HR Competencies. According to Table 4.13, a significant and positive regression equation was found (F(1, 88) = 8.809, p< .01), with an R2 of .091. Employees’ perceived HR Effectiveness increased .660 unit for each unit of overall HR Competencies. In other words, ‘Overall HR Competencies’ is a statistically significant predictor of HR Effectiveness (β=.302, p<.01).
Approximately 9 % of the variance in overall HR Effectiveness can be predicted by Overall HR Competencies. Therefore, the main hypothesis (H1) was supported.
Table 4.13.
The Linear Regression Result for the Relationship between Overall HR Competencies and Overall HR Effectiveness
Unstandardized Coefficients
Standardized Coefficients
Sig. F Sig. R2
Model B Std.
Error Beta
(Constant) 2.698 .826
.302**
.002
Overall HRCompetencies .660 .223 .004 .809 .004 .091
Note. Dependent variable: HR Effectiveness
*p< .05, **p< .01, ***p< .001.
63
H1: 1: Strategic positioner is positively related to HR effectiveness. The result of regression analysis shown in Table 4.14 reported that ‘Strategic Positioner’ is a statistically positive and significant predictor of overall HR Effectiveness (β= .293, p<.01), which can be accounted for 8.6 % of the variation of overall HR Effectiveness. Therefore, the hypothesis H1:1 was confirmed.
H1: 1: Strategic positioner is positively related to HR effectiveness. The result of regression analysis shown in Table 4.14 reported that ‘Strategic Positioner’ is a statistically positive and significant predictor of overall HR Effectiveness (β= .293, p<.01), which can be accounted for 8.6 % of the variation of overall HR Effectiveness. Therefore, the hypothesis H1:1 was confirmed.