Tổng hợp biến và câu hỏi

Biến Đo lường cũ Đo lường mới Câu hỏi

Investment Intention II1 II1

II2 II2
II3 II3
II4 II4
Product Involvement PI1 PI1
PI2 PI2
PI3 PI3
PI4 PI4
Self-Efficacy SE2 SE1
SE3 SE2
SE4 SE3

SE5 SE4
Risk and uncertainty UA3 RUA1
avoidance
RA3 RUA2
UA2 RUA3
Perceived risk and PR1 PRU1
uncertainty
PR2 PRU2
PU3 PRU3
PU2 PRU4

Descriptives
 Descriptive Statistics

N Minimum Maximum Mean Std. Deviation

PI1 120 1 7 3.31 1.522

PI2 120 1 7 3.79 1.566
PI3 120 1 7 3.31 1.704
PI4 120 1 7 3.98 1.695
SE1 120 1 7 3.63 1.414
SE2 120 1 7 3.76 1.517
SE3 120 1 7 3.80 1.412
SE4 120 1 7 3.74 1.344
RUA2 120 1 7 5.20 1.493
RUA3 120 1 7 4.92 1.601
RUA1 120 1 7 5.16 1.501
PRU1 120 1 7 4.33 1.567
PRU2 120 1 7 3.56 1.425
PRU4 120 1 7 4.58 1.559
PRU3 120 1 7 4.32 1.461
II1 120 1 7 2.65 1.776
II2 120 1 7 2.87 1.758
II3 120 1 7 2.68 1.725
II4 120 1 7 3.08 1.696
Valid N (listwise) 120

First thing first, we will take the Cronbach Alpha Cronbach’s alpha is a measure
of internal consistency, that is, how closely related a set of items are as a group. 
It is considered to be a measure of scale reliability. Cronbach's alpha is the most
common measure of internal consistency ("reliability"). It is most commonly used when you
have multiple Likert questions in a survey/questionnaire that form a scale and you wish to
determine if the scale is reliable. We need to take the Cronbach Alpha before the EFA to test
the raliabiltity.

Bước 1: Crobach Alpha

- Biến II
Reliability Statistics

Cronbach's
Alpha N of Items

.876 4

Item-Total Statistics
 Cronbach's
Scale Mean if Scale Variance Corrected Item- Alpha if Item
Item Deleted if Item Deleted Total Correlation Deleted

II1 8.62 20.438 .727 .844

II2 8.41 19.605 .807 .811
II3 8.59 20.193 .780 .822
II4 8.20 22.380 .623 .882

The alpha coefficient for the four items is .876, suggesting that the items have
relatively high internal consistency.  (A reliability coefficient of .6 or higher is
considered “acceptable” in most social science research situations.)

The first column indicates the observed variables to be included to verify the
scale reliability. The 2nd column is Corrected Item Total Correction. The values
in this column must be greater than or equal to 0.3, new observation variables
contribute to building the reliability of the scale, if the observation variable has a
relative value variable in total less than 0.3 then we will remove it.

As previously mentioned, the more the Cronbach alpha value is, the more
reliable the scale is. Should in the event that the occurrence of an observation
variable with a CAFD value greater than the Cronbach Alpha coefficient of the
initial scale (in the table Reliability Statistic) means that if the observer is
abandoned, the scale willi ncrease the reliability. If corrected Item Total
Correction of variables is less than 0.3, we should leave that variable. With the
results here have a value greater than 0.3, we will retain it.
We will apply the same procedure with others variable and we will have the
following results.

- Biến PI
Reliability Statistics

Cronbach's
Alpha N of Items

.803 4
 Item-Total Statistics

Cronbach's
Scale Mean if Scale Variance Corrected Item- Alpha if Item
Item Deleted if Item Deleted Total Correlation Deleted

PI1 11.08 16.884 .586 .769

PI2 10.60 16.612 .585 .769
PI3 11.08 14.296 .724 .698
PI4 10.41 15.807 .582 .772

The alpha coefficient for the four items is .803, suggesting that the items have
relatively high internal consistency. The corrected Item – Total Correclation >0.3
and The Cronbach Alpha If Item Deleted < Cronbach Alpha

- Biến SE

Reliability Statistics

Cronbach's
Alpha N of Items

.795 4

Item-Total Statistics

Cronbach's
Scale Mean if Scale Variance Corrected Item- Alpha if Item
Item Deleted if Item Deleted Total Correlation Deleted

SE1 11.30 11.624 .670 .712

SE2 11.18 11.255 .641 .727
SE3 11.13 12.671 .539 .777
SE4 11.19 12.711 .580 .757

- Biến RUA

Reliability Statistics
 Cronbach's
Alpha N of Items

.809 3

Item-Total Statistics

Cronbach's
Scale Mean if Scale Variance Corrected Item- Alpha if Item
Item Deleted if Item Deleted Total Correlation Deleted

RUA2 10.08 7.532 .675 .721

RUA1 10.12 7.331 .703 .692
RUA3 10.36 7.476 .600 .801

- Biến PRU

Reliability Statistics

Cronbach's
Alpha N of Items

.739 4

Item-Total Statistics

Cronbach's
Scale Mean if Scale Variance Corrected Item- Alpha if Item
Item Deleted if Item Deleted Total Correlation Deleted

PRU1 12.46 11.813 .559 .663

PRU2 13.23 13.067 .504 .694
PRU4 12.20 12.229 .516 .688
PRU3 12.47 12.520 .545 .671

Bước 2: Phân tích EFA

EFA stands for Exploratory Factor Analysis. Factor analysis can be used to find meaningful patterns
within a large amount of data. It’s possible that you will find that a certain group of questions seem to
cluster together. Exploratory factor analysis is a statistical technique that is used to
reduce data to a smaller set of summary variables and to explore the underlying
theoretical structure of the phenomena.  It is used to identify the structure of the
relationship between the variable and the respondent. 

With the reliability test of the Cronbach Alpha scale, we evaluate the relationship
between variables in the same group, the same factor, rather than considering the
relationship between all of the observed variables in other factors. In the meanwhile,
EFA considers the relationship between variables in all groups (different factors) to
detect the variables that upload the multiple factors or the misrepresentation of the
element from the beginning.

As a data analyst, the goal of a factor analysis is to reduce the number of

variables to explain and to interpret the results. This can be accomplished in two
steps:

1. factor extraction
2. factor rotation

With this analysis, we only care about 3 mains factors:

- KMO and Bartlett’s Test:

- Total Variance Explained
- Pattern Matrix

KMO and Bartlett's Test

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .761

Bartlett's Test of Sphericity Approx. Chi-Square 996.801

df 171

Sig. .000

Communalities

Initial Extraction

II1 .654 .649

II2 .732 .790
II3 .673 .752
II4 .446 .453
PRU4 .374 .397
PRU3 .458 .476
PRU1 .505 .485
PRU2 .470 .519
RUA3 .481 .515
RUA1 .642 .740
RUA2 .552 .558
SE1 .553 .695
SE2 .557 .579
SE3 .418 .399
SE4 .471 .457
PI1 .472 .495
PI2 .435 .484
PI3 .635 .716
PI4 .519 .543

Extraction Method: Principal Axis

Factoring.

Total Variance Explained

Rotation Sums of
Squared
Initial Eigenvalues Extraction Sums of Squared Loadings Loadingsa

Factor Total % of Variance Cumulative % Total % of Variance Cumulative % Total

1 4.586 24.139 24.139 4.214 22.181 22.181 3.631

2 3.447 18.141 42.280 2.996 15.766 37.947 3.014
3 1.941 10.217 52.497 1.521 8.003 45.950 2.649
4 1.572 8.275 60.772 1.083 5.700 51.651 2.452
5 1.305 6.869 67.641 .888 4.676 56.327 2.335
6 .808 4.252 71.893
7 .717 3.772 75.664
8 .677 3.564 79.228
9 .637 3.351 82.579
10 .554 2.918 85.497
11 .431 2.267 87.763
12 .414 2.177 89.941
13 .365 1.922 91.863
14 .362 1.907 93.770
15 .314 1.652 95.423
16 .274 1.444 96.866
17 .227 1.195 98.062
18 .204 1.076 99.137
19 .164 .863 100.000

Extraction Method: Principal Axis Factoring.

a. When factors are correlated, sums of squared loadings cannot be added to obtain a total variance.

Pattern Matrixa

Factor

1 2 3 4 5

II3 .873
II2 .833
II1 .762
II4 .757
PI3 .757
PI2 .751
PI1 .703
PI4 .641
SE1 .873
SE2 .700
SE4 .648
SE3 .592
RUA1 .819
RUA2 .730
RUA3 .713
PRU2 .718
PRU3 .669
PRU1 .621
PRU4 .544

Extraction Method: Principal Axis Factoring.

Rotation Method: Promax with Kaiser Normalization.
a. Rotation converged in 6 iterations.

The KMO (Kaiser-Meyer-Olkin) coefficient is an indicator for the appropriate consideration of

the factor analysis. The KMO value must reach a value of 0.5 or higher (0.5 < = KMO < = 1) is
sufficient condition for the factor analysis to be appropriate. If this value is less than 0.5, then the
factor analysis is likely not appropriate for the research dataset.
Bartlett's Test of Sphericity is used to consider the variables observed in the factor that correlated
with each other. The condition required to apply the factor analysis are observation variables that
reflect different aspects of the same factor must be correlated with each other. Therefore, if
auditing for master is not statistically significant, it is not recommended to apply the factor
analysis for the variables under consideration. Bartlett's Test has a statistical significance of <
0.05 that demonstrates observations that correlated with each other in the factor  The Bartlett’s
Test value needs to be less than 0.05.

The Eigenvalue number is a common use criterion for determining the number of factors in EFA
analysis. With this criterion, the factors which have Eigenvalue greater than or equal to 1 are
retained in the model.

Total Variance Explained greater than or equal to 50% for teachers of EFA models is
appropriate. The value of the variable is 100%, which has been showing how many% of the
elements were condensed and how much losses were lost of the observed variables.
Coefficient of loading (Factor Loading) greater than or equal to 0.5. Factor Loading: This value
represents the correlation relationship between the observed variable with the factors. The higher
the coefficient of Factor Loading, which means the correlation between the observer variable and
the larger the factor, and opposite.
According to Hair & ctg(2009,116), Multivariate Data Analysus, 7th Edtion, we have:
- Factor Loading = +- 0.3 : Minimum conditions for the observations to be retained
- Factor Loading at +- 0.5: The observation variable has a good statistical significance\
- Factor Loading at +- 0.7: The observation variable has a very good statistical significance
In fact, it is common to take loading factors 0.5 as standard with sample size from 120 to below
350. We will therefore take 0.5
Bước 3: Phân tích CFA

After finishing EFA, we have Pattern Matrix, we will use this table to conduct CFA (Confirmatory Factor
Analysis). Confirmatory factor analysis (CFA) is a multivariate statistical procedure that is used to test
how well the measured variables represent the number of constructs. Confirmatory factor analysis
(CFA) and exploratory factor analysis (EFA) are similar techniques, but in exploratory factor analysis
(EFA), data is simply explored and provides information about the numbers of factors required to
represent the data. In exploratory factor analysis, all measured variables are related to every latent
variable. But in confirmatory factor analysis (CFA), researchers can specify the number of factors
required in the data and which measured variable is related to which latent variable. Confirmatory
factor analysis (CFA) is a tool that is used to confirm or reject the measurement theory. Confirmatory
factor analysis has become established as an important analysis tool for many areas of the social
and behavioral sciences.
CMIN

Model NPAR CMIN DF P CMIN/DF

14
Default model 48 207.259 .000 1.460
2
Saturated model 190 .000 0
Independence 1060.67 17
19 .000 6.203
model 9 1

RMR, GFI

Model RMR GFI AGFI PGFI

Default model .171 .856 .807 .640
Saturated model .000 1.000
Independence
.658 .431 .367 .388
model

Baseline Comparisons
 RFI TLI
NFI IFI
Model rho rho CFI
Delta1 Delta2
1 2
Default model .805 .765 .929 .912 .927
Saturated model 1.000 1.000 1.000
Independence
.000 .000 .000 .000 .000
model

RMSEA

LO
Model RMSEA HI 90 PCLOSE
90
Default model .062 .043 .080 .141
Independence
.209 .197 .221 .000
model

CR AVE MSV ASV RUA II PI SE PRU

RUA 0.814 0.595 0.192 0.090 0.772        
II 0.878 0.646 0.299 0.137 -0.295 0.804      
PI 0.802 0.510 0.299 0.110 -0.125 0.547 0.714    
SE 0.798 0.500 0.118 0.072 0.259 0.344 0.198 0.707  
PRU 0.737 0.413 0.192 0.095 0.438 -0.204 -0.291 0.249 0.643

According to Hu & Bentler (1999). Cutoff Criteria for fit Indexes in covariance structure analysis:
Conventional criteria versus new alternatives, structural Equation Modeling, we have criteria có evaluate
Model Fit, include:

- CMIN/DF <= 3 is good, CMI/DF <= 5 is acceptable

- CFI >= 0.9 is good, CFI >= 0.95 is very good, CFI >= 0.8 is acceptable
- GFI >= 0.9 is good, GFI >= 0.95 is very good
- RMSEA <= 0.06 is good, RMSEA <= 0.08 is acceptable
- PCLOSE >= 0.05 is good, PCCLOSE >= 0.01 is acceptable

Bước 4: SEM

Structural equation modeling is a multivariate statistical analysis technique that is used

to analyze structural relationships.  This technique is the combination of factor
analysis and multiple regression analysis, and it is used to analyze the structural
relationship between measured variables and latent constructs.  This method is
preferred by the researcher because it estimates the multiple and interrelated
dependence in a single analysis. Structural equation modeling is also called causal
modeling because it tests the proposed causal relationships. 

We have 3 main results here include: Regression Weights, Standardized Regression Weights,
Squared Multiple Correlations.

Regression Weights: (Group number 1 - Default model)

Estimate S.E. C.R. P Label

.
PRU <--- SE .158 1.782 .075
089
RU .
PRU <--- .215 2.675 .007
A 080
.
PRU <--- PI -.149 -2.386 .017
062
.
II <--- PI .522 5.129 ***
102
.
II <--- PRU -.099 -.525 .599
189
II3 <--- II 1.000
.
II2 <--- II 1.073 11.920 ***
090
.
II1 <--- II .960 10.193 ***
094
.
II4 <--- II .756 7.790 ***
097
PI3 <--- PI 1.000
.
PI2 <--- PI .607 6.561 ***
092
.
PI1 <--- PI .644 7.249 ***
089
.
PI4 <--- PI .737 7.473 ***
099
SE1 <--- SE 1.000
.
SE2 <--- SE 1.041 7.359 ***
141
.
SE4 <--- SE .788 6.502 ***
121
.
SE3 <--- SE .769 6.055 ***
127
 Estimate S.E. C.R. P Label
RUA RU
<--- 1.000
1 A
RUA RU .
<--- .918 7.774 ***
2 A 118
RUA RU .
<--- .844 6.992 ***
3 A 121
PRU2 <--- PRU 1.000
.
PRU3 <--- PRU 1.194 4.765 ***
251
.
PRU1 <--- PRU 1.397 4.945 ***
282
.
PRU4 <--- PRU 1.253 4.723 ***
265
With the first table, Regression Weights, we will look at P (sig) and Estimates. By establishing a
95% confidence interval, Other variable all have sig (P) equal 0, therefore
these correlation are significant. PI, PRU has effect on II. SE, RUA, PI has
effects on PRU

Standardized Regression Weights: (Group number 1 - Default model)

Estimate
PRU <--- SE .221
PRU <--- RUA .343
PRU <--- PI -.286
II <--- PI .546
II <--- PRU -.054
II3 <--- II .852
II2 <--- II .897
II1 <--- II .794
II4 <--- II .655
PI3 <--- PI .901
PI2 <--- PI .595
PI1 <--- PI .650
PI4 <--- PI .668
SE1 <--- SE .788
SE2 <--- SE .765
SE4 <--- SE .653
SE3 <--- SE .607
RUA
<--- RUA .850
1
 Estimate
RUA
<--- RUA .784
2
RUA
<--- RUA .672
3
PRU2 <--- PRU .560
PRU3 <--- PRU .652
PRU1 <--- PRU .711
PRU4 <--- PRU .641
Continuing in consideration to the universal standardized Regression Weights
table, this is the normalized recovery coefficient table. We will rely on the
regression factor in this table. In the turn that impacts PRU, the order of
variables impacts descending as follows: SE, RUA, PI. In the turn that impacts
II, the order of variables impacts descending as follows: PI, PRU

Squared Multiple Correlations: (Group number 1 - Default model)

Estimate
PRU .290
II .318
PRU4 .411
PRU1 .505
PRU3 .425
PRU2 .313
RUA
.452
3
RUA
.615
2
RUA
.722
1
SE3 .368
SE4 .426
SE2 .585
SE1 .621
PI4 .446
PI1 .422
PI2 .354
PI3 .812
II4 .429
II1 .631
II2 .804
 Estimate
II3 .726

Standardized Indirect Effects (Group number 1 - Default model)

RUA SE PI PRU II
PRU .000 .000 .000 .000 .000
II -.018 -.012 .015 .000 .000
-.18
PRU4 .220 .142 .000 .000
4
-.20
PRU1 .244 .157 .000 .000
3
-.18
PRU3 .224 .144 .000 .000
7
-.16
PRU2 .192 .124 .000 .000
0
RUA
.000 .000 .000 .000 .000
3
RUA
.000 .000 .000 .000 .000
2
RUA
.000 .000 .000 .000 .000
1
SE3 .000 .000 .000 .000 .000
SE4 .000 .000 .000 .000 .000
SE2 .000 .000 .000 .000 .000
SE1 .000 .000 .000 .000 .000
PI4 .000 .000 .000 .000 .000
PI1 .000 .000 .000 .000 .000
PI2 .000 .000 .000 .000 .000
PI3 .000 .000 .000 .000 .000
II4 -.012 -.008 .367 -.035 .000
II1 -.015 -.009 .446 -.043 .000
II2 -.017 -.011 .503 -.048 .000
II3 -.016 -.010 .478 -.046 .000

In the end we consider the Squared Multiple Correlations table. This table represents the average R-
value of the impact of the independent variables on the dependent variables. The R squared value of
PRU is 0.29 = 29%, so the independent variables impact to 29% of the PRU variation. Similarly, The R
squared value of II is 0.318 = 31.8%, so the independent variables impact to 31.8% of the II variation