In this chapter, the hypotheses were tested by using hierarchical regression analysis on SPSS and the research questions are answered by showing statistical data.
Descriptive Statistics of Task Technology Fit
To answer the research question 1, do characteristics of S Company’s e-learning system fit those of agents’ tasks, the descriptive statistics shown in Table 4.1 may be evidence for answering it. The items of content dimension, as shown in Table 4.1, have average score at 4 which can be considered that the learners perceived the contents of e-learning courses are designed in fitness to their tasks. However, it is noted that item 4 (mean= 3.96) and 6 (mean=
3.98) are rated lower than 4. The result of item 4 shows the possibility that individuals perceive “detail” differently. For item 6, it has relatively lower value than item 5 that indicates the same idea by asking the degree that learners perceive accuracy of information.
The different mean scores may have resulted from the unclear definition of the word
“accurate” in the questions.
In technology and presentation dimensions, the average ratings are obviously lower than that of content dimension. Item 12 in technology dimension is especially low (mean= 3.73) which might imply that the operational functions and commands on the e-learning system could be improved. In addition, the software and hardware stability as well as system update are perceived poorer which are both rated at 3.78.
58
Table 4.1. Descriptive Statistics－Question Items of TTF Descriptive Statistics－Question Items of TTF
Dimensions Items Mean Std.
Deviation Content 1 The e-learning courses provide critical information
that would be very useful to me in my job.
4.17 .681
2 The information maintained by the e-learning courses is exactly what I need to carry out my tasks.
4.21 .669
3 The e-learning courses provide useful information and increase my job efficiency.
4.10 .790
4 The e-learning courses maintain information at an appropriate level of detail for my purposes.
3.96 .807
5 The information that I use or would like to use is accurate enough for my purposes.
4.19 .697
6 The information I use or need is accurate enough. 3.98 .761 7 The e-learning courses maintain consistent
information throughout.
4.17 .681
8 The e-learning courses maintain information that is consistent with other sources of information.
4.22 .692
Technology 9 It is easy to locate the e-learning courses on
corporate portal. 4.03 .860
10 I can find specific topics in the e-learning courses
quickly and easily when I need it. 3.93 .862 11 It is easy to change the selection and format of
information made available by our e-learning systems. (personal settings: font styles or the amount of charts or text)
4.07 .865
12 The instructions of all operational functions and
commands on the e-learning system are clear. 3.73 .916 13 The instructions on how to operate the e-learning
system is easy to find. 4.00 .766
14 I can get the help I need in accessing and
understanding the information. 3.93 .830
15 It is easy to get assistance when I am having trouble
finding or using information. 4.00 .783
16 It is easy to learn how to use the e-learning system. 4.07 .731 17 The e-learning system is convenient and easy to use. 4.00 .833
(table continues)
Table 4.1. (continued)
Dimension Items Mean Std.
Deviation 18 The e-learning system including software and
hardware works stably. 3.78 .979
19 I can count on the system to be “up” and available
when I need it. 3.78 .999
20 I can get information that is current enough to meet
my needs. 3.99 .770
Presentation 21 The information is up-to-date enough that reflects
the current market. 3.77 .875
22 The information in the e-learning courses is
organized in a way that is easy to understand. 4.02 .753 23 The information is presented in a readable and
useful format. 3.96 .832
Note. N=151
Generally speaking, the overall average score of task technology fit is 3.99 above the medium 3 of the 5 point scale (shown in Table 4.2). Therefore, we can assume that TTF in S Company is above average and the research question is positively answered. Nevertheless, since average scores of technology and presentation dimensions do not exceed point 4, it shows that there are improvement opportunities in the e-learning courses which S Company should be attentive to.
60
Correlations
Means, standard deviations, and correlations for each variable are shown in Table 4.2.
As expected, all three major study variables have positive correlations with each other.
Specifically, TTF has a positive correlation with learning outcome and training transfer. On the other hand, learning style with a high mean of 8.44 does not show any significant association with any of the study variables. The study also found that gender and e-learning experience are correlated with TTF, and that age and job experience are correlated with training transfer. This study involves training transfer scores of the second month and third month.
Table 4.2. Correlation Analysis Results Correlation Analysis Results
Variables Mean St.Dev 1 2 3 4 5 6 7 8 9 10 11
1. Age 2.54 .68
2. Gender .67 .47 .056
3. Education 2.06 .26 .216** -.041
4. Major 2.52 1.29 .034 -.037 -.033
5. Job Experience 2.27 1.12 .620** -.130 -.041 .135
6. Sales Experience 1.31 .63 .482** .110 .008 .028 .450**
7. Marriage .94 .26 -.335** -.092 .147 -.143 -.230** -.286**
8. E-learning Experience
2.95 1.19 .152 -.091 .093 .111 .145 .142 -.220**
9. Learning Style 8.44 2.05 .017 -.036 .086 -.005 .006 .016 .135 .071
10. TTF 3.99 .60 -.027 .162* -.065 .063 .008 .085 -.071 .197* .037 (0.965)
11. Learning Outcome
65.79 12.50 -.016 -.075 -.014 -.014 .094 .023 -.117 .058 -.029 .276**
12. Training Transfer (2nd month)
75.05 8.91 .259** .014 .040 .118 .217** .116 -.138 .116 -.016 .185* .278**
13. Training Transfer (3rd month)
74.99 8.46 .288** .010 .034 .081 .234** .102 -.105 .134 .020 .228** .288**
Note. * p <0.05, ** p <0.01, ***p <0.001
62
Regression for Hypothesis Testing
Hierarchical regression was used in statistical analysis. Demographic factors such as age, gender, job experience, and e-learning experience were entered into regression equations as control variables. Three regression equations were constructed to test the study hypotheses.
The first hypothesis test is shown as Table 4.3, with “learning outcome” as the dependent variable and the four control variables and scores of TTF as the independent variables. Model 1 shows that none of the control variables are significant and the model only have a 2.2%
(R2= .022) explanatory power; while after adding TTF as independent variable (as shown in Model 2), the result shows that TTF has significant impact on learning outcome (β= .292, p<0.01) and R2 raises to .100 with ∆R2 significant at 0.1% level.
Table 4.3. Results of Regression Analysis of Hypothesis 1 Results of Regression Analysis of Hypothesis 1
Independent Variable Learning Outcomes
Model 1 Model 2
(Constant)
Age -.110 -.075
Gender -.044 -.102
Job Experience .148 .126
E-learning Experience .050 -.015
TTF .292**
R2 .022 .100
Adjusted R2 -.005 .069
∆R2 .078**
F .814 3.231**
N 151 151
Note. Standardized regression coefficients are shown.
* p <0.05, ** p <0.01, ***p <0.001
Table 4.4 shows the results of hypothesis 2 in which “training transfer” is the dependent variable and the four control variables and TTF again as the independent variables. In Model 1, age is significant at 5% level and all control variables have 9.7% explanatory power (R2= .097). Also, Model 2 also confirmed the significance of age on the process of training transfer (β= .248, p<0.05), while the impact of TTF on training transfer is significant at 1%
level (β= .233, p<0.01). The R2 increased to .146 and the R2 change is significant at 1% level as shown in Model 2 in Table 4.4.
Table 4.4. Results of Regression Analysis of Hypothesis 2 (3rd months) Results of Regression Analysis of Hypothesis 2 (3rd month)
Independent Variable Training Transfer
Model 1 Model 2
(Constant)
Age .220* .248*
Gender .012 -.035
Job Experience .089 .071
E-learning Experience .091 .039
TTF .233**
R2 .097 .146
Adjusted R2 .072 .117
∆R2 .050**
F 3.900** 4.970***
N 151 151
Note. Standardized regression coefficients are shown.
* p <0.05, ** p <0.01, ***p <0.001
The third hypothesis is shown as Table 4.5 with “training transfer” as the dependent variable again and learning outcome is additionally added to the equation of hypothesis 2.
The result of Model 3 shows that “learning outcomes” is significant at 1% level (β= .240, p<0.01), and while significant in Model 2, the coefficient of TTF has dropped from .233
64
which is significant at 1% level to .163. Although it is still significant at 5% level, it only reaches the significance boundary. Therefore, the result shows that learning outcome has a partial to full mediating effect on the relationship between TTF and training transfer.
Furthermore, hypothesis 3 has 19.8% explanatory power with significant R2 change at 1%
level.
Table 4.5. Results of Regression Analyses of Hypothesis 3 (3rd months) Results of Regression Analyses of Hypothesis 3 (3rd month)
Independent Variable Training Transfer
Model 1 Model 2 Model 3
(Constant)
Age .220* .248* .266**
Gender .012 -.035 -.010
Job Experience .089 .071 .041
E-learning Experience .091 .039 .043
TTF .233** .163*
Learning Outcomes .240**
R2 .097 .146 .198
Adjusted R2 .072 .117 .165
∆R2 .050** .052**
F 3.900** 4.970*** 5.929***
N 151 151 151
Note. Standardized regression coefficients are shown.
* p <0.05, ** p <0.01, ***p <0.001
From Table 4.3, the regression shows that variable “TTF” is significant at .001. This implies that employees’ perceived task technology fit can lead to high learning outcomes. As shown in Table 4.4, TTF is also significant at 1% level. Therefore, hypothesis 1 and 2 are supported.
Hypothesis 1: Perception of task technology fit of an e-learning course has a positive influence on employees’ learning outcome.
Hypothesis 2: Perception of task technology fit of an e-learning course has a positive influence on employees’ training transfer.
From Table 4.5, after adding learning outcomes into the regression equation as independent variables, the originally significant effect of task technology fit on training transfer has dropped, which is evident that learning outcome has a partial to full mediating effect on the relationship between TTF and training transfer. Therefore, hypothesis 3 is also supported.
Hypothesis 3: Employees’ learning outcome mediates the relationship between task technology fit of an e-learning course and training transfer.
To be noted, the regression equation for hypothesis 1 has an R² of .100; therefore, it can only explain 10% of the variance in learning outcomes. The regression model for hypothesis 2 has an R² of 0.146; therefore, it can only explain 14.6% of the variance in training transfer.
With both TTF and learning outcome in the equation, the model for hypothesis 3 achieves a much higher R² of .198, which explains 19.8% of the variance in training transfer.
As training transfer scores are recorded each month in S Company, this study also tests the effects of TTF on second month training transfer. Table 4.6 shows the regression analyses of hypotheses 2 and 3 with the second month training transfer scores for observing the longitudinal change of TTF effects. The results show similar pattern as the third month test.
Model 2 in Table 4.6 supports hypothesis 2 showing TTF significant at 0.05 level with 12.1%
explanatory power. When adding learning outcome into the equation, Model 3 shows that TTF is completely insignificant. Therefore, hypothesis 3 is also supported. Comparing the data to 2nd month training transfer result, the 3rd month training transfer result has higher R2 in both models of hypothesis 2 and 3. The higher explanatory power shows the fact that the effect of training transfer may take longer time to observe, usually between three to six
66
months.
Table 4.6. Results of Regression Analyses of Hypotheses 2, 3 (2nd months) Results of Regression Analyses of Hypotheses 2, 3 (2nd month)
Independent Variable Training Transfer
Model 1 Model 2 Model 3
(Constant)
Age .215* .234* .245*
Gender .016 -.022 .000
Job Experience .095 .082 .051
E-learning Experience .070 .029 .035
TTF .181* .117
Learning Outcomes .218**
R2 .091 .121 .164
Adjusted R2 .066 .091 .129
∆R2 .030* .043**
F 3.625** 3.951** 4.646***
N 151 151 151
Note. Standardized regression coefficients are shown.
* p <0.05, ** p <0.01, ***p <0.001
As a general perception, age, gender, employees’ job and e-learning experience may have positive influence on learning effectiveness. While in this case, only age shows significant relationship with training transfer. Table 4.7 further tests the moderating effect of learning style between TTF and learning outcome, though it has been revealed that learning style has no correlations with any variables in Table 4.2. As shown in Model 3 of Table 4.7, learning style shows no moderating effects on learning outcome with β= -.023 which is not significant at all. Therefore, hypothesis 4 is not supported.
Hypothesis 4: Employees’ learning style will moderate the relationship between task technology fit of an e-learning course and learning outcome.
Table 4.7. Results of Regression Analyses of Hypothesis 4 Results of Regression Analyses of Hypothesis 4
Independent Variable Learning Outcome
Model 1 Model 2 Model 3
(Constant)
Age -.110 -.074 -.075
Gender -.044 -.104 -.102
Job Experience .148 .125 .123
E-learning Experience .050 -.012 -.010
TTF .293*** .294**
Learning Styles (LS) -.043 -.044
LS*TTF -.023
R2 .022 .102 .103
Adjusted R2 -.005 .065 .059
∆R2 .080** .001
F .814 2.727* 2.334*
N 151 151 151
Note. Standardized regression coefficients are shown.
* p <0.05, ** p <0.01, ***p <0.001
As it is supported in literature that learning style may influence learning outcome, this study seeks to explain the insignificant result by examining the distribution of learners’
preference in learning styles. The researcher calculates the frequency of the employees’
scores and found out a homogeneous learning preference of the sample. According to ISL (Felder & Soloman, 2000), scores ranged from 10 to 11 represent a strong preference, 8 to 9 represent a moderate preference, and 6 to 7 represent a mild preference. The distribution of visual-verbal learning preference is shown in Table 4.8.
68
Table 4.8. Descriptive Statistics—Distribution of Visual-verbal Preference Descriptive Statistics—Distribution of Visual-verbal Preference
Preference to Visual Learning Scores Frequency
Strong Preference 10-11 52
Moderate Preference 8-9 60
Mild Preference 6-7 25
Little/None Preference below 6 14
Note. N=151
Table 4.8 shows that most scores fall between 8 to 11 which stands for strong or moderate preference to visual learning preference. In that case, the standard deviation is small at 2.05 and the average score is at 8.44. Therefore, the homogeneous learning style may result in the insignificant moderating effect that rejects hypothesis 4.