Professional Documents
Culture Documents
Dravni univerzitet
Novi Pazar, Serbia
ospirtovic@np.ac.rs
I.
INTRODUCING
PREVIOUS RESEARCH
Period of time
using the services
of telecom
operators
Telecom operator
RESPONDENTS CHARACTERISTICS OF
POPULATION SAMPLE
Category
Male
Female
Less than 20 years
From 20 to 24 years
From 25 to 34 years
More than 34 years
Less than 6 months
From 7 to 12 months
From 13 to 24 months
From 25 to 36 months
More then 36 months
Eronet
BH Telecom
M:tel
Frequency
472
267
110
294
259
76
10
77
224
289
139
122
357
260
%
63,9
36.1
14,9
39,8
35,0
10,3
1,4
10,4
30,3
39,1
18,8
16,5
48,3
35,2
RESULTS OF RESEARCH
Construct
Switching intention
Quality of service
Price
Customer satisfaction
Question
mark
p22 - p24
p16 - p21
p11 - p15
p7 - p10
Cronbach
alpha
0,845
0,915
0,892
0,853
B. SEM Analysis
Structural Equation Modeling (SEM) is a multivariate
statistical technique that is a combination of confirmatory
factor analysis (CFA), path analysis (PA) and regression
analysis and applied in the analysis of hypothetical
relationships between endogenous (variables that can not
be directly measured) and exogenous (variables which can
be directly measured) variables. Using AMOS statistical
software (which is used for the modeling and simulation
of SEM, SEM statistical analysis and data processing as
well as the graphical representation of the results of SEM
analysis) we conducted testing of ours hypothesis [3].
Figure 3. shows the diagram of the model for measuring
customer switching intent through the quality and price of
service with the calculated parameters through AMOS
program, where by labels e1-e24 represent the residual
variance (formed under the influence of uncontrolled
factors) exogenous variables p1-p24, while mark e25, e26
and e27 represent residual variance of endogenous
variables [7]. (Appendix B)
Based on the SEM analysis first sought to ascertain the
degree of appropriateness of the proposed model with
empirical model. In Table IV., it was given an overview of
goodness-of-fit indices that measure the degree of
suitability.
TABLE IV.
Goodness-of-fit Indices
Benchmark
Value
2,736
.90
0,932
.90
0,917
.08
0,025
.05
0,049
.90
0,944
.90
0,963
.90
0,959
.50
0,767
.50
0,862
.50
0,845
Error
of
In this study, as shown in Table IV., all goodness-offit indices are better then their corresponding
recommended values. Values obtained goodness-of-fit
indices indicate that the level of appropriateness defined
model data is satisfactory, ie. there is overlap empirical
and target matrix, and model is acceptable for further
analysis.
In this study, as shown in Table V., correlation
between construct shows that all variables are positively
correlated with each other. Construct switching intention
and customer satisfaction have shown the greatest degree
of correlation 0.812. Construct switching intention and
price have shown the least degree of correlation 0.603.
The resulting correlation coefficients suggest that the
hypothesis H1 to H4 can be considered confirmed. All
correlation coefficients are positive. (Appendix C)
In this study, as shown in Table VI., standardized
structural coefficients that estimate the direct causal
relations between the latent variables specified in the
previously defined structural model. It can be seen that the
p-value for each correlation is less than or equal to 0.001
(*** p < 0.001), which means that there is a statistically
significant difference between these constructs. Also, it
can be seen in Table VI. that the estimated coefficient of
construct satisfaction is 0.928 (has the strongest impact on
coustomer switching intent) and the estimated coefficient
of construct quality of service is 0.480 (has the strongest
impact on customer satisfaction). Estimated coefficient of
construct price is 0.386. This means that construct quality
of service and construct price (indirectly) and construct
satisfaction (directly) have pozitive influence on customer
switching intent. Thus, we can conclude that customer
switching intent can be achieved directly from customer
satisfaction and indirectly from quality of service and
price. (Appendix D)
The resulting standardized structural coefficients
indicate that the hypothesis H1 to H4 can be considered
confirmed. All structural coefficients are statistically
significant and have the expected direction.
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
VI.
CONCLUSION
[10]
[11]
[12]
[13]
[14]
APPENDIX A
QS4p19
QS5p20
QS6p21
CL1p1
CL2p2
CL3p3
CL4p4
CL5p5
CL6p6
CS1p7
CS2p8
CS3p9
CS4p10
P1p11
P2p12
P3p13
P4p14
P5p15
QS1p16
QS2 p17
QS3p18
SI1p22
SI2p23
SI3p24
Agree
Strongly Agree
Measurement items
Uncertain
Mark
Disagree
TABLE I.
Strongly disagree
APPENDIX B
Fig 3. AMOS diagram model for measuring customer switching intent through the quality and price of service with the calculated parameters
APPENDIX C
TABLE V.
Quality of Service
Price
Customer satisfaction
Customer loyalty
Switching intention
Quality
of Service
1,000
,645
,791
,622
,642
Customer
satisfaction
Customer
loyalty
Switching
intention
1,000
,787
,812
1,000
,639
1,000
1,000
,743
,584
,603
APPENDIX D
TABLE VI.
Hypothesis
Path
Estimate
S.E.
C.R.
p-value
Result
H3
Customer satisfaction
<---
Price
0,386
0,039
9,913
***
Accepted
H1
Customer satisfaction
<---
Quality of Service
0,480
0,038
12,524
***
Accepted
H4
Customer loyalty
<---
Customer satisfaction
0,818
0,046
17,731
***
Accepted
H2
Switching intention
<---
Customer satisfaction
0,928
0,052
17,788
***
Accepted