36
Figure 4.2. CFA of Endogenous Construct
4.3.2. Full Structural Equation Model Analysis
4.3.2.1. Model Fit
After passing the CFA for both exogenous and endogenous constructs, the full model must be tested to make sure that the model was fit.
Comparing to the same eight criterias and cut-off values used in the previous CFA, the result of model fit test was reported in table 4.7. Unfortunately
even had passed the CFA, the full model was found not fit since the score of three criterias were not good as expected, they are the chi-square, the
significance probability, and the AGFI. Based on the model evaluation that was not fit, checking the validity of all instruments must be done. Indicators
37 with loading factor that less than 0.5 was not valid and must be cut. The
loading factor of all indicators used could be seen in table 4.8 and based on it, there were actually four indicators that were not valid B1, C1, P4, and
M4. After cut the four unvalid indicators, the model fit test be done again and the result could be seen in figure 4.4 while the model evaluation could
be seen in table 4.8 that by a modification, the model was fit.
Table 4.7. Full Model Structural Goodness of Fit Indices
before Modification Goodness of Fit
Indices Cut Off Value
Result Model
Evaluation
Chi-Square Expected to be low
172.92 Bad
Significance Probability
≥ 0.05 0.00
Bad RMSEA
≤ 0.08 0.05
Good CMINDF
≤ 2.00 1.57
Good GFI
≥ 0.90 0.91
Good AGFI
≥ 0.90 0.88
Marginal TLI
≥ 0.90 0.94
Good CFI
≥ 0.95 0.95
Good
Source: Primary Data, 2014
Table 4.8. Squarred Multiple Correlations
Estimate
C1 .178
B4 .781
M4 .009
P4 .013
B3 .703
B2 .638
B1 .013
38
Estimate
C4
.631
C3 .651
C2 .615
M1
.536
M2 .655
M3 .657
P1 .825
P2 .644
P3 .745
Figure 4.3. Full Model Structural before Modification
39
Figure 4.4. Full Model Structural after Modification
40
Table 4.9. Full Model Structural Goodness of Fit Indices
after Modification Goodness of Fit Indices
Cut Off Value Result
Model Evaluation
Chi-Square Expected to be low
58.88 Good
Significance Probability ≥ 0.05
0.37 Good
RMSEA ≤ 0.08
0.02 Good
CMINDF ≤ 2.00
0.15 Good
GFI ≥ 0.90
0.96 Good
AGFI ≥ 0.90
0.93 Good
TLI ≥ 0.90
0.99 Good
CFI ≥ 0.95
0.99 Good
Source: Primary Data, 2014
4.3.2.2. Reliability
Before performing the hypotheses testing, it must be checked that all indicators that had passed the validity test also passed the reliability test.
This must be done to make sure that all indicators used here, including the borrowing habit indicators were reliable and could be employed in the future.
The reliability test here done by calculating and evaluating the construct reliability score.
The formula to calculate the construct reliability was:
Construct Reliability =
standardized loading
2
standardized loading
2
+ εj
After calculating the construct reliability of each variable, they must be compared to the cut-off valued that should be greater than 0.7. The result
41 of the calculation was reported in table 4.9 that the lowest score was only
0.8, shown by money retention. Therefore all twelve indicators had passed the reliability test and proved as reliable indicators.
Table 4.10. Construct Reliability
Estimate Construct Reliability
P3 --- PEER_ACCEPTANCE .865
0.939 P2 --- PEER_ACCEPTANCE
.803 P1 --- PEER_ACCEPTANCE
.907 M3 --- MONEY_RETENTION
.812 0.896
M2 --- MONEY_RETENTION .810
M1 --- MONEY_RETENTION .732
C4 --- COMPULSIVE_BUYING .805
0.904 C3 --- COMPULSIVE_BUYING
.803 C2 --- COMPULSIVE_BUYING
.784 B2 --- BORROWING_HABIT
.801 0.903
B3 --- BORROWING_HABIT .840
B4 --- BORROWING_HABIT .881
Source: appendix 12
4.4. Statistic Descriptive
There were total twelve indicators that proved as valid and reliable indicators. Therefore only the mean and standard deviation of those twelve
indicators that reported in table 4.10. Peer acceptance consisted of three indicators, as well as the other constructs. In general, all statements of peer
acceptance showed a low tendency of peer acceptance with the total average of 2.45. This score indicated that the students did not try too hard to impress
42 their peers and to get their peers‟ acceptance. The total average of money
retention came as the highest score of all with the score 3.54, it was in the interval of 3.41-4.20. The meaning of this score was that the students were
capable on managing their pocket money, they saved their extra money, and they did not easily spend their pocket money. In contrast, the total average of
CBB was low since it was only 2.31. The meaning of this score was that the students did not have the tendency as compulsive buyers. The last total score
reported on table 4.10 was the score of borrowing habit, which was 2.41. Similar to the tendency of peer acceptance and CBB, the students also
showed a low tendency of borrowing habit. They did not easily decide to borrow money from their friends even from their closest friends and
relatives.
Table 4.11. Descriptive Statistic of Peer Acceptance, Money Retention, CBB,
and Borrowing Habit Statement
Mean Standard
Deviation Peer Acceptance
I use trendy products. 2.23
0.86 I use high quality products in order to be
accepted by my friends. 2.58
0.79 I often follow my friends to buy produts.
2.27 0.85
Total Average 2.45
Money Retention
I don‟t spend money easily. 3.51
0.83 I am capable of managing my pocket money
3.51 0.83
43 so that I seldom experience lack of money.
In times when I have extra money, I save it. 3.61
0.82 Total Average
3.54
CBB
I often have a strong urge and spontaneous desire to go and buy something.
2.31 0.76
I find it isn‟t hard to restrain my self not to shop.
2.31 0.81
For me, shopping is a way of facing the stress
2.32 0.84
Total Average 2.31
Borrowing Habit
I try as much as possible not to borrow money.
2.39 0.93
I will think carefully before deciding to borrow money.
2.39 0.92
I am not used to borrow money when I need it.
2.46 0.94
Total Average 2.41
Source: appendix 13 Note: reverse statement
The answers’ interval categories: 1.00
-1.80 very low; 1.81- 2.60 low; 2.61-3.40 medium; 3.41-4.20 high; 4.21-5.00 very
high.
4.5. Hypothesis Testing