Factor Analysis and Structural Equation Modeling
The data for part b) and c) are to be found at my homepage:
http://www.bi.no/users/fgl96027/
a) Figure 1 shows a structural model. Substitute standard notation for the letters A
through G and a through w. Use the standard notation to represent the model in
equation form. Classify the parameters as belonging to either the measurement
part or the structural part of the model.
m
x1
a
l
k
A
n
E
x2
b
o
C
D
w
G
c
x3
F
s
t
u
v
B
p
q
y1
y2
g
h
y3
y4
i
j
r
x4
x5
x6
d
e
f
Figure 1
b) Use the dataset in the PSF-file: Case-Gra6020S-2004.psf
i)
Start with a data screening (N=533)
ii)
Do an exploratory factor analysis and try to find a suitable factor
structure
iii)
Based on ii), estimate and test the CFA-model. It could be a good
idea to try several estimators. Comment on your findings.
c) In a study to determine the predictors of drinking and driving behavior among 18to 24-years-old males, the following model was proposed (N=356):
1   111   12 2   13 3   1
 2   211   2
1 : Intentions to drink and drive ( y1 and y2 )
 2 : Drinking and driving behavior ( y3 and y 4 )
1 : Attitude toward drinking and driving ( x1  x3 )
 2 : Social norms pertaining to drinking and driving ( x4  x6 )
 3 : Perceived control over drinking and driving ( x7 and x8 )
Use the covariance matrix to estimate the parameters of the structural model.
Comment on the model fit.
The covariance matrix is to be found in the file: Drinkd.cov. Here you will find a
matrix of 12 variables listed in the following order y1  y 4 , x1  x8 . I.e., there are four
y-variables and eight x-variables, and the y-variables are listed first in the covariance
matrix.