Scribd
Upload
113 views5 pages

Assignment 6 Multivariate Statistical Analysis (Pls-Sem) : Fatin Najlaa Binti Abd Rahman (2020941359)

This document reports on the results of a statistical analysis using PLS-SEM. It includes measurement model results showing acceptable indicator loadings, composite reliability, and average variance extracted. Discriminant validity was also confirmed. The analysis found a positive effect of perceived usefulness on intention to use and a positive but weaker effect of statistical software self-efficacy on intention to use. Model prediction was better using LM than PLS-SEM. A second analysis found customer expectation positively impacts customer loyalty and satisfaction, and satisfaction positively impacts loyalty, with all indirect effects being statistically significant, indicating mediation.

Uploaded by

najlanizam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
113 views5 pages

Assignment 6 Multivariate Statistical Analysis (Pls-Sem) : Fatin Najlaa Binti Abd Rahman (2020941359)

This document reports on the results of a statistical analysis using PLS-SEM. It includes measurement model results showing acceptable indicator loadings, composite reliability, and average variance extracted. Discriminant validity was also confirmed. The analysis found a positive effect of perceived usefulness on intention to use and a positive but weaker effect of statistical software self-efficacy on intention to use. Model prediction was better using LM than PLS-SEM. A second analysis found customer expectation positively impacts customer loyalty and satisfaction, and satisfaction positively impacts loyalty, with all indirect effects being statistically significant, indicating mediation.

Uploaded by

najlanizam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

FATIN NAJLAA BINTI ABD RAHMAN (2020941359)

ASSIGNMENT 6
MULTIVARIATE STATISTICAL ANALYSIS (PLS-SEM)

Measurement Model Table


Construct Item Loading CR AVE Cronbach's
Alpha

Perceived PU1 0.849 0.909 0.714 0.867


Usefulness
PU2 0.856
PU3 0.866
PU4 0.809
Statistical SSSE2 0.589 0.791 0.573 0.823
Software
SSSE3 0.995
Self SSSE4 0.617
Efficacy
Intention to BI1 0.859 0.902 0.755 0.838
Use
BI2 0.899
BI3 0.848

The measurement model was assessed by examining the loadings, average


variance extracted (AVE) and composite reliability (CR) as suggested by Hair et al.
(1998) and Ramayah et al. (2018). More specifically, the fulfilment of following three

ASSIGNMENT 6&7 1
FATIN NAJLAA BINTI ABD RAHMAN (2020941359)

criteria was assessed: all indicator loadings should exceed 0.5, the AVE for each
construct should be greater than 0.5 and the CR should exceed 0.7. As shown in
Measurement Model Table, that all the indicator loadings are above 0. 5, the CR
values range from 0.79 to 0.91, and the AVE ranges from 0.57 to 0.76. All three
conditions for reliability and convergent validity of the measures thus hold.

Cronbach alpha values for Perceived Usefulness is 0.87, Statistical Software Self
Efficacy is 0.82 and Intention to Use is 0.84 which is above .8 is considered reliable.

Discriminant Validity Table

  1 2 3

INTENTION - - -
TO USE

PERCEIVED 0.468 - -
USEFULNES
S
STATISTICAL 0.104 0.271 -
SOFTWARE
SELF
EFFICACY

To assess the discriminant validity, the more recent criterion called


Heterotraitmonotrait ratio (HTMT) suggested by Henseler et al. (2015) was used,
instead of the usual Fornell and Larcker’s (1981) criterion. If the HTMT ration is less
than 0.85, then the constructs are distinct. As shown in Discriminant Validity Table,
all the ratios are less than 0.85, thus, discriminant validity was also confirmed.

VIF Table
  PERCEIVED STATISTICAL
USEFULNESS SOFTWARE SELF

ASSIGNMENT 6&7 2
FATIN NAJLAA BINTI ABD RAHMAN (2020941359)

EFFICACY
VIF 1.057 1.057

All the variables are regressed against a common variable and if the variance
inflation factor (VIF) is less than or equals 5, then, there is no multicollinearity
problem. As shown in VIF Table the analysis yielded a VIF less than 5 and,
therefore, no multicollinearity problem is this study.

Path Coefficients
    Std.Bet Std. t- p- BCI LL BCI UL f2
a Error value value
H1 PERCEIVED 0.398 0.042 9.371 0 0.324 0.464 0.179
USEFULNESS
-> INTENTION

H2 STATISTICAL 0.025 0.068 0.372 0.355 -0.174 0.093 0.001


SOFTWARE
SELF
EFFICACY ->
INTENTION

Perceived Usefulness had a positive effect on Intention to Use (β 0.398, p < 0.01)
and Statistical Software Self Efficacy had a positive effect on Inention to use (β
0.025, p < 0.01).

PLS Predict
PLS-
SEM -
PLS-SEM LM
LM
  RMSE
Q²_predic
  RMSE t RMSE  
BI1 1.189 0.118 1.18 0.009
BI2 1.254 0.092 1.232 0.022
BI3 1.213 0.14 1.197 0.016

PLS-SEM analysis produces higher prediction errors for all the indicators as PLS-
SEM analysis has higher RMSE value compare to LM.
ASSIGNMENT 7

MEDIATION AND MODERATION TESTING


ASSIGNMENT 6&7 3
FATIN NAJLAA BINTI ABD RAHMAN (2020941359)

Measurement Model Table

Construct Item Loading CR AVE

CUSTOMER CUEX1 0.805 0.797 0.663


EXPECTATION CUEX2 0.823    
CUSTOMER CUSL1 0.828 0.867 0.766
LOYALTY CUSL3 0.92    

CUSTOMER CUSA 1 0.811 0.871 0.693


SATISFACTIO CUSA2 0.84    
N CUSA3 0.847    

The measurement model was assessed by examining the loadings, average


variance extracted (AVE) and composite reliability (CR) as suggested by Hair et al.
(1998) and Ramayah et al. (2018). More specifically, the fulfilment of following three
criteria was assessed: all indicator loadings should exceed 0.5, the AVE for each
construct should be greater than 0.5 and the CR should exceed 0.7. As shown in
Measurement Model Table, that all the indicator loadings are above 0. 5, the CR

ASSIGNMENT 6&7 4
FATIN NAJLAA BINTI ABD RAHMAN (2020941359)

values range from 0.79 to 0.87, and the AVE ranges from 0.66 to 0.76. All three
conditions for reliability and convergent validity of the measures thus hold.

Path Coefficients
Std t p BCI
  Std Beta Error value value LL BCI UL
CUSTOMER
EXPECTATION -> 0.20
CUSTOMER LOYALTY 0.282 0.045 6.236 0 1 0.364
CUSTOMER
EXPECTATION ->
CUSTOMER
SATISFACTION 0.455 0.054 8.455 0    
CUSTOMER
SATISFACTION ->
CUSTOMER LOYALTY 0.619 0.068 9.174 0    

Customer Expectation had a positive effect on Customer Loyalty (β 0.282, p < 0.01),
Customer Expectation had a positive effect on Customer Satisfaction (β 0.455, p <
0.01) and Customer Satisfaction had a positive effect on Customer Loyalty (β 0.619,
p < 0.01).

The bootstrapping analysis has shown that all three indirect effect, β=0.282, β=0.455
and β=0.619, are significant which t-value of 6.236, 8.455 and 9.174. The indirect
effect 95% Boot CI Bias Corrected: [LL=0.201, UL= 0.364) do not straddle a o in
between indicating there is mediation (Preacher and Hayes,2004,2008). Thus, we
can conclude that the mediation effect are statistically significant.

ASSIGNMENT 6&7 5

You might also like