Set up Kenny-Judd (1984) example model in LISREL 8--"constrained" approach

DA NI=9 NO=500 MA=CM

!  reading in covariance matrix--must analyze covariances for interactions
CM FI=KJEXAMPL.CM
ME
0 0 0 0 0 0 0 0 5.0

!  labels for measures of 4 factors:  X and Z (main), XZ (interaction),
! Y (dependent)
LA           
x1 x2 z1 z2 x1_z1 x1_z2 x2_z1 x2_z2 y 

!  selecting the product of the "1" measures
!  and the product of the "2" measures 
SE
y    x1 x2    z1 z2    x1_z1            x2_z2  /     

MO NY=1 NE=1 NX=6 NK=3 PS=SY PH=SY GA=FU,FR C
             LX=FU,FI LY=FU,FI TD=SY TE=SY,FI KA=FI AL=FR TY=FI

! construct labels
LE
Y
LK
X   Z   XZ

!  Here's the main effects measurement model
VA 1.0 LY(1,1) LX(1,1) LX(3,2)
FR LX(2,1) LX(4,2)

!   Here are the constraints on PHI
FI PH(3,1) PH(3,2)   
!  after mean centering and with normality, interaction factor should be
!  orthogonal

CO PH(3,3) = PH(1,1) * PH(2,2) + PH(2,1)^2

!  Here is Marsh et al's value for the mean of the interaction construct
CO KA(3) = PH(2,1)

!  Here are the loadings for the interaction measures
VA 1.0 LX(5,3)
CO LX(6,3) = LX(2,1) * LX(4,2)

!  Here are the constraints on the measurement error variances for the interaction measures.
CO TD(5,5) = PH(1,1) * TD(3,3) + PH(2,2) * TD(1,1) + TD(1,1) * TD(3,3)
CO TD(6,6) = PH(1,1) * TD(4,4) * LX(2,1)^2 + PH(2,2) * TD(2,2) * LX(4,2)^2      C
                          + TD(2,2) * TD(4,4)

OU AD=OFF RS MI