Analysing
Validity and Reliability of
instrument items
DR CHUA YAN PIAW (UM)
Prof Dr Chua Yan Piaw
University of Malaya
1
3
2
5
3
RESEARCH STATISTICS Example 2
A researcher conducts a study to examine
the effect of Teamwork on Performance.
Using data file Data Example 2, draw the
research model, analyse and report the
results of PLS-SEM analysis for:
1. Construct validity
2. Construct reliability
DR CHUA YAN PIAW (UM)
3. Multicollinearity of items in each variable
of the instrument items
2) Report the results of analysis.
Convergent validity (construct validity)
RESEARCH STATISTICS
1. Loadings ≥ 0.70
Remove indicators that have factor
loadings < .50
DR CHUA YAN PIAW (UM)
4
RESEARCH STATISTICS
Convergent validity
2. AVE ≥ 0.50
Note: If AVE smaller than 0.50, remove
items that have smallest loadings one by
one, to increase AVE.
DR CHUA YAN PIAW (UM)
Result: Convergent validity achieved.
Construct reliability ≥ 0.70
RESEARCH STATISTICS
1. Cronbach's alpha
2. Composite reliability (a and c)
DR CHUA YAN PIAW (UM)
Result: Construct reliability achieved.
10
5
RESEARCH STATISTICS
Discriminant validity:
1. Cross loadings
DR CHUA YAN PIAW (UM)
11
RESEARCH STATISTICS
Discriminant validity:
2 Fornell-Larcker criterion
DR CHUA YAN PIAW (UM)
12
6
RESEARCH STATISTICS
Discriminant validity:
3. HTMT < 0.90
DR CHUA YAN PIAW (UM)
13
Discriminant validity:
RESEARCH STATISTICS
Solving problem of discriminant validity
Remove items that has high inter-
correlation values with other variables.
DR CHUA YAN PIAW (UM)
14
7
RESEARCH STATISTICS
Discriminant validity:
After removing Teamwork item 1
DR CHUA YAN PIAW (UM)
Discriminant validity achieved
Result: Construct validity achieved.
15
RESEARCH STATISTICS
Multicollinearity
VIF < 0.50
DR CHUA YAN PIAW (UM)
Result: Multicollinearity does not
occur among items in each
measurement model
16
8
RESEARCH STATISTICS Results:
Item 2 of Teamwork was removed due to
small loading values (< 0.70).
Item 1 of Teamwork was removed due to
discriminant validity. This is because it
correlated highly with the four items of
Performance.
Teamwork has 3 valid and reliability items:
DR CHUA YAN PIAW (UM)
Items 3, 4, and 5.
Performance has 4 valid and reliability
items: items 1, 2, 3 and 4.
17
Example 3
RESEARCH STATISTICS
A researcher conducts a study to examine the
effects of Attitude on Thinking and Performance.
Using data file Data Example 3, draw the
multivariate model in SmartPLS, analyse and
report the results of PLS-SEM analysis for:
a. loadings of indicators of the two measurement
models.
b. Path coefficient
DR CHUA YAN PIAW (UM)
c. R-square
d. validity, reliability and multicollinearity of the
instrument items of the three variables
2) Report the results of analysis.
18
9
10
19
20
RESEARCH STATISTICS DR CHUA YAN PIAW (UM) RESEARCH STATISTICS DR CHUA YAN PIAW (UM)
