This section presents the findings about the respondents profile in terms of their age, gender, level of education and monthly income. The data are shown in frequencies and percentage. The respondents have been classified into six groups of age: less than 18 years old; between 19-25 years old; between 26-35 years old; between 36-45 years old; between 46-55 years old; 56 years old and above. Table 4.1 indicates that there were no respondents whose age was less than 18 years old. Around one-fifth (33 or 21.6 percent) of the respondents whose age was between 19-25 years old. More than half (78 or 51.0 percent) of the respondents whose age was between 26-35 years old. About (33 or 21.6 percent) of the respondents whose age was between 36-45 years old. Only (8 or 5.2 percent) respondents whose age was between 46-55 years old and only one person whose age was 56 years old and above.
What is your age group?
Frequency (n)
Percent (%)
Less than 18 years old
0
0.0
19 - 25 years old
33
21.6
26 - 35 years old
78
51.0
36 - 45 years old
33
21.6
46 - 55 years old
8
5.2
Above 56 years old
1
0.7
Total
153
100.0
Table 4.1: distribution of respondents according to their ages
Table 4.2 shows that there were (42 or 27.5 percent) of female respondents while the male dominated by (111 or 72.5 percent) respondents.
What is your gender?
Frequency (n)
Percent (%)
Female
42
27.5
Male
111
72.5
Total
153
100.0
Table 4.2: distribution of respondents according to their gender
The overall distribution of age and gender is shown in Figure 4.1 below:
Figure 4.1: distributions of respondents according to their ages & gender
4.1.2 Level of education
The respondents have been classified into six groups of educational level: high school, diploma (two years), bachelor's degree, master's degree, doctoral degree and other.
Table 4.3 indicates that there were almost (11 or 7.2 percent) of the respondents who had high school. About (5 or 3.3 percent) of the respondents who had diploma (two years). About one-third (52 or 34.0 percent) of the respondents who had bachelor's degree. More than third (60 or 39.2 percent) of the respondents who had master's degree. About (23 or 15 percent) of the respondents who had doctoral degree and about (2 or 1.3 percent) holding other types of certificates.
What is your education level?
Frequency (n)
Percent (%)
High School
11
7.2
Diploma (two years)
5
3.3
Bachelor's Degree
52
34.0
Master's Degree
60
39.2
Doctoral Degree
23
15.0
Other
2
1.3
Total
153
100.0
Table 4.3: distribution of respondents according to their educational level
Figure 4.2: distributions of respondents according to their educational level
4.1.3 Monthly Income
The respondents have been classified into seven groups of monthly income: Less than 2,999; 3,000 - 5,999; 6,000 - 8,999; 9,000 - 11,999; 12,000 - 14,999; 15,000 - 19,999 and More than 20,000 SAR per month.
Table 4.4 shows that there were almost (12 or 7.8 percent) of the respondents who had less than 2,999 SAR income. About (15 or 9.8 percent) had between 3K - 5,999 SAR. About (22 or 14.4 percent) had between 6K - 8,999 SAR. About 17.6 and 17.0 percent had less than 12K and 15K respectively. The majority (28 or 18.3 percent) of respondents had between 15K - 19,999 SAR per month and about (23 or 15.0 percent) have had more than 20K SAR as monthly income.
What is your monthly income (SAR)
Frequency (n)
Percent (%)
Less than 2,999
12
7.8
3,000 - 5,999
15
9.8
6,000 - 8,999
22
14.4
9,000 - 11,999
27
17.6
12,000 - 14,999
26
17.0
15,000 - 19,999
28
18.3
More than 20,000
23
15.0
Total
153
100.0
Table 4.4: distribution of respondents according to their monthly income
Figure 4.3: distributions of respondents according to their monthly income
4.2 Level of Selected Variables
This part discusses the respondents' level of agreement on system quality factors, information quality factors and service quality factors. The findings are presented in frequencies, percentages, and means. The discussion also emphasizes the data sufficiency and variables effect on e-Commerce growth within Saudi Arabia.
4.2.1 Level of agreement based on system quality
In terms of System Quality Approach, it can be seen in Table 4.5 that 58 (37.9 percent) of the respondents are "Strongly Agree", while 29 (19 percent) of the respondents are "Agree" and 28 (18.3 percent) "Somewhat Agree". Twenty-two (14.4 percent) of the respondents are "Not Sure". Ten (6.5 percent) are "Somewhat Disagree" while 2 (1.3 percent) of them are "Disagree". Only 4 respondents (2.6 percent) are "Strongly Disagree".
Range of Mean
Scores
Verbal Description
Frequency
Percentage
1.0 - 1.86
Strongly Disagree
4
2.6 %
1.87 - 2.73
Disagree
2
1.3 %
2.74 - 3.60
Somewhat Disagree
10
6.5 %
3.61 - 4.47
Not Sure
22
14.4 %
4.48 - 5.34
Somewhat Agree
28
18.3 %
5.35 - 6.21
Agree
29
19 %
6.22 and above
Strongly Agree
58
37.9 %
TOTAL
153
100 %
Overall Mean = 5.45 ("Agree")
Table 4.5: distribution of respondents according to the level of agreement on system quality
The findings indicated factor 2 (Ease of Use) as the highest level of agreement among the respondents (M=5.66, SD=1.531) in this category. On the other hand, factor 4 (Reliability) had the lowest level of agreement among the respondents (M=5.34, SD=1.717).
System Quality Factors
Mean
Std. Deviation
Verbal Description
Availability
5.56
1.674
Agree
Ease of use
5.66
1.531
Agree
Elasticity
5.39
1.667
Agree
Reliability
5.34
1.717
Somewhat Agree
Responsiveness
5.34
1.686
Somewhat Agree
Security
5.42
1.834
Agree
Overall Mean = 5.45 ("Agree")
Table 4.6: Means, Std. Dev., and Verbal Description for system quality (n=153)
Figure 4.4: means distributions of system quality factors
In the case of the factor analysis, one important aspect is to test the assumptions. The two key techniques used are the Kaiser-Meyer-Olkin (KMO) sampling adequacy test and the Bartlett test for sphericity. The KMO tests the appropriateness of the data, while the Bartlett tests for correlations. For system quality, these tests are shown on Table 4.7.
KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy.
.898
Bartlett's Test of Sphericity
Approx. Chi-Square
703.067
df
15
Sig.
.000
Table 4.7: KMO and Bartlett tests for System Quality
As shown on Table 4.7, the KMO is 0.898. According to Field (2005, p650), the recommended minimum KMO is 0.5 Values between 0.5 and 0.7 are considered as mediocre. KMO of values between 0.7 and 0.8 are considered as good, while values above 0.8 are considered as great. Based on the KMO of 0.898 produced in this analysis, it has been justified that the factor analysis was appropriate for this data.
Bartlett's measure tests the null hypothesis that the original correlation matrix is an identity matrix. For factor analysis to work, it in necessary for some variables to have relationships; if the R-matrix were an identity, then all correlation coefficients would be zero. Hence there is a need to test for significance (have p <0.05). A large result indicates that R-matrix is not an identity matrix. For regulatory requirement, Bartlett's test is highly significant (p = 0.000), indicating that the factor analysis was appropriate for this data.
The results of the more robust factor analysis techniques for system quality show that a single solution explained about 75% of the variance as shown on Table 4.8 below.
Total Variance Explained
Component
Initial Eigenvalues
Extraction Sums of Squared Loadings
Total
% of Variance
Cumulative %
Total
% of Variance
Cumulative %
dimension0
1
4.472
74.527
74.527
4.472
74.527
74.527
2
.463
7.725
82.252
3
.365
6.087
88.339
4
.279
4.656
92.996
5
.252
4.192
97.188
6
.169
2.812
100.000
Extraction Method: Principal Component Analysis.
Table 4.8: Variance explained by the system quality factors
4.2.2 Level of agreement based on information quality
In terms of Information Quality Approach, it can be seen in Table 4.7 that 46 (30.1 percent) of the respondents are "Strongly Agree", while 36 (23.5 percent) of the respondents are "Agree" and 39 (25.5 percent) "Somewhat Agree". Seventeen (11.1 percent) of the respondents are "Not Sure". Nine (5.9 percent) are "Somewhat Disagree" while 2 (1.3 percent) of them are "Disagree". Only 4 respondents (2.6 percent) are "Strongly Disagree".
Range of Mean
Scores
Verbal Description
Frequency
Percentage
1.0 - 1.86
Strongly Disagree
4
2.6 %
1.87 - 2.73
Disagree
2
1.3 %
2.74 - 3.60
Somewhat Disagree
9
5.9 %
3.61 - 4.47
Not Sure
17
11.1 %
4.48 - 5.34
Somewhat Agree
39
25.5 %
5.35 - 6.21
Agree
36
23.5 %
6.22 and above
Strongly Agree
46
30.1 %
TOTAL
153
100 %
Overall Mean = 5.38 ("Agree")
Table 4.9: distribution of respondents according to the level of agreement on information quality
The findings indicated factor 6 (Simplicity) as the most influential factor amongst the respondents (M=5.64, SD=1.431) in this category. On the other hand, factor 4 (Personalization) had the lowest level of agreement among the respondents (M=5.16, SD=1.506).
Information Quality Factors
Mean
Std. Deviation
Verbal Description
Accuracy
5.58
1.621
Agree
Completeness
5.47
1.577
Agree
Dynamic Content
5.21
1.596
Somewhat Agree
Personalization
5.16
1.506
Somewhat Agree
Relevance
5.24
1.504
Somewhat Agree
Simplicity
5.64
1.431
Agree
Overall Mean = 5.38 ("Agree")
Table 4.10: Means, Std. Dev., and Verbal Description for information quality (n=153)
Figure 4.5: means distributions of information quality factors
The factor analysis produced a single factor solution with an explanatory variance of 78% (Table 4.11). The KMO was 0.913 and Bartlett test (p = 0.000), indicating the appropriateness of the factor analysis.
Total Variance Explained
Component
Initial Eigenvalues
Extraction Sums of Squared Loadings
Total
% of Variance
Cumulative %
Total
% of Variance
Cumulative %
dimension0
1
4.671
77.857
77.857
4.671
77.857
77.857
2
.370
6.171
84.028
3
.336
5.599
89.627
4
.270
4.500
94.127
5
.187
3.112
97.239
6
.166
2.761
100.000
Extraction Method: Principal Component Analysis.
Table 4.11: Variance explained by the information quality factors
4.2.3 Level of agreement based on service quality
In terms of Service Quality Approach, it can be seen in Table 4.9 that 46 (30.1 percent) of the respondents are "Strongly Agree", while 36 (23.5 percent) of the respondents are "Agree" and 39 (25.5 percent) "Somewhat Agree". Seventeen (11.1 percent) of the respondents are "Not Sure". Nine (5.9 percent) are "Somewhat Disagree" while 2 (1.3 percent) of them are "Disagree". Only 4 respondents (2.6 percent) are "Strongly Disagree".
Range of Mean
Scores
Verbal Description
Frequency
Percentage
1.0 - 1.86
Strongly Disagree
1
0.7 %
1.87 - 2.73
Disagree
5
3.3 %
2.74 - 3.60
Somewhat Disagree
13
8.5 %
3.61 - 4.47
Not Sure
23
15 %
4.48 - 5.34
Somewhat Agree
40
26.1 %
5.35 - 6.21
Agree
33
21.6 %
6.22 and above
Strongly Agree
38
24.8 %
TOTAL
153
100 %
Table 4.11: distribution of respondents according to the level of agreement on service quality Overall Mean = 5.19 ("Somewhat Agree")
The findings indicated factor 2 (Understanding) as the highest level of agreement among the respondents (M=5.32, SD=1.370) in this category. On the other hand, factor 5 (Dedication) had the lowest level of agreement among the respondents (M=4.96, SD=1.589).
Service Quality Factors
Mean
Std. Deviation
Verbal Description
Commitment
5.24
1.551
Somewhat Agree
Understanding
5.32
1.370
Somewhat Agree
Knowledge
5.03
1.560
Somewhat Agree
Privacy Protection
5.25
1.710
Somewhat Agree
Dedication
4.96
1.589
Somewhat Agree
Human Factor
5.31
1.599
Somewhat Agree
Overall Mean = 5.19 ("Somewhat Agree")
Table 4.12: Means, Std. Dev., and Verbal Description for service quality (n=153)
Figure 4.6: means distributions of service quality factors
Again, the factor analysis produced a single factor solution with an explanatory variance of 67% (Table 4.13). The KMO was 0.900 and Bartlett test (p = 0.000), indicating the appropriateness of the factor analysis.
Total Variance Explained
Component
Initial Eigenvalues
Extraction Sums of Squared Loadings
Total
% of Variance
Cumulative %
Total
% of Variance
Cumulative %
dimension0
1
4.017
66.947
66.947
4.017
66.947
66.947
2
.582
9.706
76.652
3
.486
8.102
84.755
4
.366
6.099
90.854
5
.305
5.077
95.931
6
.244
4.069
100.000
Extraction Method: Principal Component Analysis.
Table 4.13: Variance explained by the service quality factors
4.3 Reliability Analysis Test
Reliability can be defined as the degree to which an experiment, test, or measuring procedure would produce the same result on repeated trials (Writing guides, 2009). Furthermore, reliability could be defined as the degree to which measures are free from error and thus yield consistent results (Zikmund, 2003).
There are several different types of reliability coefficients such as Split half reliability, Guttman, Parallel, Strictly parallel and Cronbach's alpha. Cronbach's alpha is one of the most commonly used measures because it can be interpreted as a correlation coefficient and it ranges in value from 0 to 1 (Coakes and Steed, 2003). Hence, Cronbach's alpha was used as a measurement of reliability for each variable in this study.
Variables
No. of measures
Cronbach's Alpha
System Quality
6
0.931
Information Quality
6
0.943
Service Quality
6
0.898
Table 4.14: the results of reliability test
From the analysis done on the instruments listed under each variable in the questionnaire, Table 4.14 shows that Cronsbach's Alpha for the variables system quality, information quality, and service quality are 0.931, 0.943, and 0.898 respectively. The internal consistency reliability of the measures can be considered as "great" as it achieved more than 0.8 Alpha values (Field, 2005, p. 668).
4.4 Demographic effect on the dimensions' factors
As stated in chapter three, the effect of demographic differences is to be examined wither it effect the significance of factors in each dimension or not. The following hypothesis was built to test this issue:
H0: Demographic differences have no impact on variables' significance in an IS model.
H1: The importance of success variables in an IS model vary due to demographic differences.
To cautiously investigate this issue, a comparison was made between cases with different demographic (from same category) in each on of the three dimensions (i.e. System Quality, Information Quality, and Service Quality) to clearly mark any differences that might exist. A detailed list of all the mean comparison made is included in appendix B.
4.4.1 Age difference effect on system quality
To measure the influence of age on system quality, several comparisons have been made to investigate the case. In general, all tests showed an increase in system quality demanding along with the increase in age.
The ease of use and system flexibility features' significance increases with respect to the increase in age as descried in Table 4.14. "Youth are careless & risk takers, while elders are more conservative & risk averse" this statement can be clearly observed when monitoring (Reliability & Security) which increases with the growth of age.
Consequently, the effect of age difference on system quality could be undoubtedly seen in this matter.
H0 is rejected ïƒ research hypothesis H1 is true
Case Summaries
What is your age group?
Availability
Ease of use
Elasticity
Reliability
Responsiveness
Security
19 - 25 years old
N
33
33
33
33
33
33
Mean
5.36
5.48
5.15
5.03
5.06
4.97
Std. Deviation
1.800
1.523
1.584
1.776
1.802
1.960
26 - 35 years old
N
78
78
78
78
78
78
Mean
5.55
5.68
5.37
5.32
5.29
5.46
Std. Deviation
1.711
1.608
1.722
1.732
1.691
1.785
36 - 45 years old
N
33
33
33
33
33
33
Mean
5.76
5.73
5.61
5.58
5.70
5.79
Std. Deviation
1.480
1.506
1.638
1.621
1.447
1.691
46 - 55 years old
N
8
8
8
8
8
8
Mean
5.50
6.00
5.63
6.25
5.63
5.50
Std. Deviation
1.773
1.069
1.847
1.165
2.134
2.330
Above 56 years
N
1
1
1
1
1
1
Mean
5.00
5.00
5.00
2.00
4.00
4.00
Std. Deviation
.
.
.
.
.
.
Total
N
153
153
153
153
153
153
Mean
5.55
5.66
5.39
5.34
5.34
5.42
Std. Deviation
1.670
1.531
1.667
1.717
1.686
1.834
Table 4.15: age difference effect on system quality factors
Figure 4.7: age effect on means distributions of system quality
4.4.2 Level of education effect on information quality
To test the indirect effect of education level on the information quality dimension, three comparisons have been carried out.
First, both doctoral and master's equally "Agree" on the importance of information accuracy; high school level, on the other hand, are "Not Sure" about that. While diploma holders "Strongly Agree" with the importance of dynamic contents, bachelor's degree holders see less significance, thus, they tend to "Somewhat Agree" with that statement. Unlike bachelor's degree holders who "Somewhat Agree" with the importance of information simplicity, doctoral degree holders do "Agree" on its weight for information quality.
By looking at the three comparisons made, it could be clearly stated that the effect of education level is significant in this situation.
H0 is rejected ïƒ research hypothesis H1 is true.
Case Summaries
What is your education level?
Accuracy
Completeness
Dynamic Content
Personalization
Relevance
Simplicity
Bachelor's
Degree
N
52
52
52
52
52
52
Mean
5.37
5.23
5.10
5.04
4.98
5.40
Std. Deviation
1.715
1.676
1.660
1.633
1.578
1.636
Diploma
(two years)
N
5
5
5
5
5
5
Mean
6.40
6.60
6.60
6.40
6.00
6.20
Std. Deviation
1.342
.894
.894
.894
1.414
1.304
Doctoral
Degree
N
23
23
23
23
23
23
Mean
5.83
5.83
5.26
5.39
5.48
6.09
Std. Deviation
1.527
1.370
1.630
1.438
1.534
1.125
High School
N
11
11
11
11
11
11
Mean
4.36
4.73
4.27
4.55
4.27
4.91
Std. Deviation
2.335
2.102
2.240
2.018
2.149
1.921
Master's
Degree
N
60
60
60
60
60
60
Mean
5.82
5.58
5.33
5.18
5.48
5.77
Std. Deviation
1.334
1.441
1.361
1.308
1.200
1.184
Other
N
2
2
2
2
2
2
Mean
5.50
5.50
5.50
5.50
5.50
5.50
Std. Deviation
2.121
2.121
2.121
2.121
2.121
2.121
Total
N
153
153
153
153
153
153
Mean
5.58
5.47
5.21
5.16
5.24
5.64
Std. Deviation
1.621
1.577
1.596
1.506
1.504
1.431
Table 4.16: education level effect on information quality factors
Figure 4.8: education level effect on means distributions of information quality
4.4.3 Gender difference effect on service quality
Different gender has different interpretation of service quality provided by e-commerce support centres. To test the indirect effect of gender difference on the service's quality provided, a small comparison has been made.
Women tend to be more concerned about service centre willingness to help (commitment) and they "Agree" on the importance of human interaction ability in support centres. Men, on the other hand, give less attention "Somewhat Agree" to these two aspects. An interesting case to be noticed in this context is the privacy protection. In a conventional society such as Saudi Arabia, people tend to be more conservative when it comes to giving personal information. In such society, women are expected to be more conservative then men. Results, however, revealed that men were actually more sensitive toward privacy protection then women as show in Table 4.17.
Accordingly, it could be concluded that gender differences have an effect on service quality dimension as shown here.
H0 is rejected ïƒ research hypothesis H1 is true.
Case Summaries
What is your gender?
Commitment
Understanding
Knowledge
Privacy Protection
Dedication
Human Factor
Female
N
42
42
42
42
42
42
Mean
5.38
5.36
5.12
5.05
5.00
5.40
Std. Deviation
1.637
1.394
1.485
1.834
1.593
1.466
Male
N
111
111
111
111
111
111
Mean
5.18
5.31
4.99
5.32
4.95
5.27
Std. Deviation
1.521
1.367
1.593
1.663
1.595
1.651
Total
N
153
153
153
153
153
153
Mean
5.24
5.32
5.03
5.25
4.96
5.31
Std. Deviation
1.551
1.370
1.560
1.710
1.589
1.599
Table 4.17: gender difference effect on service quality factors
Figure 4.9: gender difference effect on means distributions of service quality
4.4 Summary of Findings
The study revealed a number of interesting cases that need some attention in order to understand the behavior and logical reasoning behind it; in order to help build a better e-Commerce system that is more considerate and sensitive to the needs of targeted consumers.
Some of interesting ranking information, regarding the participants, that could be found from the previous tables includes:
Age: More than half (51%) of the respondents are between 26-35 years old. This indicates the fact that Saudi Arabia is a young nation with a median age of 24.9 years (male: 26 years, female: 23.4 years) (The World Fact book, 2010).
Gender: The majority (72.5 %) of the participants are males. This is due to cultural limitation of contact between the two genders (women segregation) within the country. The minority was done via relatives.
Education: More than (88 %) of the participants are holders of bachelor, master's, or doctoral degrees. This is understandable since the online survey was circulated amongst Saudi Students in the UK & Saudi Arabia.
The findings indicate that simplicity (Ease of use) was the most pointed feature amongst the various system quality factors. A user friendly designed system with simple navigation ability seemed more important to consumers than reliability power, robustness or even security standards. Thus, e-Commerce marketing strategy should focus on sending the image of a simple & user friendly e-Commerce system rather than focusing on the facts of security or multi-access capabilities (i.e. PC & Mobile, 24/7, …etc).
As for Information quality measures, the findings indicated that simplicity was again the key factor acquired by consumers. Although completeness and accuracy were of high demands (5.47 & 5.58 means), keeping it simple, straight forward, and in understandable terms was most favorite feature. This is obvious when compared to comprehensiveness which might leads to long lists of annoying terms and conditions catalog.
The ability of e-Commerce support centers to understand consumers' specific needs was the most important factor of service quality. An argument might be that, a well committed and knowledgeable support personal would not be helpful if the consumer needs were mistakenly interpreted.
Based on the data of 153 respondents, the multi-items measures were subjected to a series of validity and reliability checks. For the multi-item scale, the set of factors that correspond to each dimension was initially subjected to an examination of Cronbach's alpha and item-to-total correlations test.
Thus, all measures appeared to be uni-dimensional, internally consistent, reliable and valid for analysis of the model. Furthermore, this chapter has examined the influence of demographic effects on the dimensions' (System, Information and Service) factors prioritization process. The relationship was conducted by computing the differences measures of the means and Standard Deviations, which supported the hypotheses that all the variables have a significance impact on e-Commerce growth and adoption in Saudi Arabia.
