TITLE: Longitudinal Model for Spuriousness using Waves 1 and 2 from Windle et al. Data
DATA: FILE is C:\wincov.txt; ! Replace C: with path to folder in which file is located
type is covariance;
nobservations=433;
VARIABLE: NAMES ARE da1 pa1 so1 in1 da2 pa2 so2 in2
da3 pa3 so3 in3 da4 pa4 so4 in4 ;
UseVAR = da1-in2;
! Using waves 1 and 2 from this dataset.
analysis:
model:
! A latent variable for each time 1 measure; all loadings fixed to one.
x1a by da1@1;
x1b by pa1@1 ;
x1c by so1@1 ;
x1d by in1@1 ;
! A latent variable for each time 2 measure; all loadings are free
! and equal k for that measure.
x2a by da2*1;
x2b by pa2*1;
x2c by so2*1;
x2d by in2*1;
! No "error variance" in the measures.
da1@0;
pa1@0;
so1@0;
in1@0;
da2@0;
pa2@0;
so2@0;
in2@0;
! Force equality of cross-sectional relationships.
x1a with x1b (a1);
x1a with x1c (a2);
x1a with x1d (a3);
x1b with x1c (a4);
x1b with x1d (a5);
x1c with x1d (a6);
x2a with x2b (a1);
x2a with x2c (a2);
x2a with x2d (a3);
x2b with x2c (a4);
x2b with x2d (a5);
x2c with x2d (a6);
! Force equality of over-time relationships.
! These records would be dropped to see if there was evidence for
! over-time causal effect,
x1a with x2b (b1);
x1b with x2a (b1);
x1a with x2c (b2);
x1c with x2a (b2);
x1a with x2d (b3);
x1d with x2a (b3);
x1b with x2c (b4);
x1c with x2b (b4);
x1b with x2d (b5);
x1d with x2b (b5);
x1c with x2d (b6);
x1d with x2c (b6);
output: tech3 sampstat stand;