The 2015 International Conference of Management Sciences ICoMS 2015, April 23, UMY, Indonesia | 97
QUAL
.11
COM COMIT
TRUST x1.1
.50
e1
1.00 1
x1.2
.38
e2
1.39 1
x1.3
.34
e3
1.64 1
x1.4
.57
e4
.82 1
x3.7
.45
e16
1.00 1
x3.6
.48
e15
1.13 1
x3.5
.32
e14
1.16 1
x3.4
.69
e13
.95 1
x3.3
.56
e12
.98 1
x3.2
.55
e11
1.11 1
x3.1
1.07
e10
1.52 1
y1.1
.32
e17
1.00 1
y1.2
.33
e18
1.12 1
y1.3
.39
e19
1.07 1
y1.4
.46
e20
1.13 1
5.02 .03
-.53 .21
.02 .72
3.50
z1
1 .13
z3
1 .18
z2
1
y1.5
.92
e21
.74 1
Chi-squares=345.671 df=171
prob=.000 GFI=.865
AGFI=.817 TLI=.898
RMSEA=.071
emp
2.78
e9
1.00 1
ass
2.02
e8
.87 1
res
2.95
e7
1.03 1
rel
3.23
e6
1.05 1
tan
4.29
e5
.64 1
.45 .10
.13 .18
-.35 -.15
-.10 -.06
.51 .05
-.18 .22
.10
COM
X1.1 .51
e1
1.00 1
X1.2 .39
e2
1.40 1
X1.3 .31
e3
1.77 1
X1.4 .56
e4
.90 1
QUAL
EMP 2.69
e9
1.00 1
ASS 1.98
e8
.87 1
RES 2.99
e7
1.03 1
REL 3.36
e6
1.04 1
TRUST COMMIT
.39
z2
.53
z3
3.81
z1
4.96 .62
-.73 1
.29 .20
1 1
TAN 4.36
e5
1 .63
.08
Chi-square=56.671 df=40
p=.042 GFI=.954
AG|FI=.924 TLI=.977
RMSEA=.046
Commitment X
2
= a0 + β
1
X
1
+ e X
3
= a1 + β
2
X
1
+ β
3
X
2
+ e Y = a2 +
β
4
X
1
+ β
5
X
2
+ β
6
X
3
+ e
4. RESEARCH METHOD
Cross section in the collection of data that will be used to test the hypothesis that six interconnected. Data
were obtained through a survey method, interviews and the literature or publications related to the topic under
study. While the sources of data used comes from the primary data and secondary data. The type of data that
is used is qualitative data ordinal data stated in an interval scale Likert scale.
Population in this research were all students UNISBANK Semarang. Sampling is done by
proportional stratified sampling. Criteria for sample strata is UNISBANK students to study levels of D3
and regular S1, D3 and transfer S1, and S2, with sample sizes ranging from 100 to 200 that generate
significant difference. Variable effectiveness of communication, trust and
Relationship Commitment was measured using a questionnaire developed by Sharma and Patterson
2000. Variable quality of service is measured by a questionnaire developed Lassar, Manolis and Winsor
2000. Analysis and testing data using SEM Structural
Equation Model which is run by the program AMOS. Hypotheses that have been developed, which later
developed into a graphical model and formed structural equation model Structural Equation Model SEM.
Use of this SEM is due to its ability to display a comprehensive model along with the ability to measure
the influence between factors theoretically exist. Therefore SEM is usually viewed as a combination of
factor analysis and regression analysis, and of course can be applied separately only in the factor analysis ie
Confirmatory Factor Analysis or only in the regression analysis.
5. ANALYSIS AND RESULTS
Respondents are students UNISBANK Semarang, the number of respondents is 202 people, who were taken
by proportional stratified sampling, with the following composition:
D3 S1
S2 Total
Number of Sample
Reguler Transfer
Reguler Transfer
24 4
104 40
From the results of data processing with AMOS program showed the following results:
Figure-2 Strucural Model
AMOS computational models based on SEM, in Table 1, the evaluation criteria of goodness - of - fit as
follows:
Table-1 Criteria of Evaluation
Goodness – Of – Fit
The above table shows that the criteria used to have 1 good value and 5 the margin, therefore, this model can
not be accepted because there are many who have a margin criteria.
To get better results, especially in the goodness of fit of the model, then made some changes to the initial
structural models. Changes were made to alter the confidence
trust and
variable connectivity
commitments commit be the observed variables. Research model changes appear in the image below:
No Criteria
Result Model
Critical Value
Estimated Model
1 Chi Square 345,671
Small Margin
2 Probability
0,000 ≥ 0,05
Margin 3
GFI 0,865
≥ 0,9 Margin
4 TLI
0,898 ≥ 0,95
Margin 5
RMSEA 0,071
≤ 0,08 Good
6 AGFI
0,817 ≥ 0,9
Margin
The 2015 International Conference of Management Sciences ICoMS 2015, April 23, UMY, Indonesia | 98
Figure-3 Structural Model Changes
AMOS computational models based on SEM, in Table 2, the evaluation criteria of goodness - of - fit as
follows: AMOS computational models based on SEM, in Table 2, the evaluation criteria of goodness - of - fit
as follows:
Table -2 Criteria of Evaluation
Goodness – Of – Fit
Table 2 above shows that the criteria used to generate 1 margin value, therefore, in general this model is
acceptable .. Thus it can be stated that this test produces a good confirmation on dimension - the
dimension of factors and relationships - a causal relationship between factors.
5.1 Hypothesis Testing
Based on
AMOS computational
hypothesis testing appears in Table 3 as follows: Table-3
Hypothesis Testing Result
Regression Weights
Causalities Relation Estimate
Std. Estimate
S.E. C.R.
P
QUAL ---
COM 4.958
0.631 1.143 4.337
TRUST ---
QUAL 0.292
0.738 0.033 8.973
TRUST ---
COM 0.199
0.064 0.268 0.744
0.457 COMMIT
--- TRUST
0.62 0.62 0.091
6.848 COMMIT
--- COM
-0.728 -0.234 0.329
-2.214 0.027
COMMIT ---
QUAL 0.084
0.211 0.045 1.863
0.062 The test results in Table 3 indicate that the pathways
are analyzed showed a significant causal relationship, but the path to TRUST COM communication
effectiveness affect the trust and QUAL path to COMMIT quality of service affect the Relationship
Commitment. It is seen from the path coefficients standardized estimate and the estimate with a CR
value greater than ± 2.0 or level of significance P test the hypothesis that less than 5. Further description of
hypothesis testing in research, described as follows: a. Communication Effectiveness influence on the
Quality of Service
The first hypothesis proposed in this study is the effectiveness of communication significant positive
effect on the quality of service. In Table 3 shows the magnitude of the path
coefficients standardized estimate and the estimate with a value of CR = 4,337 greater than ± 2.0 or
level of significance P test the hypothesis that less than 5. Therefore the proposed hypothesis is
accepted.
b. Communication Effectiveness influence on Trust
The second hypothesis proposed in this study is the effectiveness of communication significant positive
effect on trust. In Table 3 shows the magnitude of the path coefficients standardized estimate and the
estimate with a value of CR = 0744 which is less than ± 2.0 or level of significance P test the
hypothesis that greater than 5. Therefore the proposed hypothesis is not accepted.
c. Influence Communication Effectiveness against Relationship Commitment
The third hypothesis proposed in this study is the effectiveness of communication significant positive
effect on the Relationship Commitment. In Table 3 shows the magnitude of the path coefficients
standardized estimate and the estimate with a value of CR = -2214 greater than ± 2.0 or level of
significance P test the hypothesis that less than 5. Therefore the proposed hypothesis is accepted.
d. Effect of Quality of Service of the Trust
The fourth hypothesis proposed in this study is the quality of service significant positive effect on
trust .. In Table 3 shows the magnitude of the path coefficients standardized estimate and the estimate
with a value of CR = 8973 which is greater than ± 2.0 or level of significance P test the hypothesis
that less than 5. Therefore the proposed hypothesis is accepted.
e. Influence of the Quality of Service Relationship Commitment
The fifth hypothesis proposed in this study is a significant positive effect of service quality on the
Relationship Commitment. In Table 3 shows the magnitude of the path
coefficients standardized estimate and the estimate with a value of CR = 1,863 less than ± 2.0 or level
of significance P test the hypothesis that greater than 5. Therefore the proposed hypothesis is
No Criteria
Result Model Critical
Value Estim
ated Model
1 Chi Square
56.671 Small
Good 2
Probability 0,042
≥ 0,05 Margi
n 3
GFI 0,954
≥ 0,9 Good
4 TLI
0,977 ≥ 0,95
Good 5
RMSEA 0,046
≤ 0,08 Good
6 AGFI
0,924 ≥ 0,9
Good
The 2015 International Conference of Management Sciences ICoMS 2015, April 23, UMY, Indonesia | 99
accepted.
f. Effect of Commitment Confidence Connectedness