Hi all,
I am estimating a model with SEM, using the following command:
The results in matrix form is:
When I am predicting the latent variables, I get the following results:
My questions are:
1) Why the std. dev. of the predicted latent variables are smaller than the sqrt of the variances of the variables in matrix Phi?
2) If I use
the latent variables are standardized in the Phi matrix; zero mean and unit variance. Is there any way to predict standard latent variables directly, or one should always predict them first and then standardize them with:
Thanks for taking the time,
Emma
I am estimating a model with SEM, using the following command:
Code:
sem (visual -> v1, ) (visual -> v2, ) (visual -> v3, ) (verbal -> v4, ) /// (verbal -> v5, ) (verbal -> v6, ) (speed -> v7, ) (speed -> v8, ) /// (speed -> v9, ), covstruct(_lexogenous, diagonal) vce(sbentler) /// standardized latent(visual verbal speed ) cov( visual*verbal visual*speed /// verbal*speed) nocapslatent
PHP Code:
Endogenous variables on endogenous variables
| observed
Beta | v1 v2 v3 v4 v5 v6 v7 v8 v9
-------------+---------------------------------------------------------------------------------------------------
observed |
v1 | 0
v2 | 0 0
v3 | 0 0 0
v4 | 0 0 0 0
v5 | 0 0 0 0 0
v6 | 0 0 0 0 0 0
v7 | 0 0 0 0 0 0 0
v8 | 0 0 0 0 0 0 0 0
v9 | 0 0 0 0 0 0 0 0 0
-----------------------------------------------------------------------------------------------------------------
Exogenous variables on endogenous variables
| latent
Gamma | visual verbal speed
-------------+---------------------------------
observed |
v1 | 1 0 0
v2 | .4907708 0 0
v3 | 1.23306 0 0
v4 | 0 1 0
v5 | 0 1.319723 0
v6 | 0 2.247793 0
v7 | 0 0 1
v8 | 0 0 1.065948
v9 | 0 0 1.655825
-----------------------------------------------
Covariances of error variables
| observed
Psi | e.v1 e.v2 e.v3 e.v4 e.v5 e.v6 e.v7 e.v8 e.v9
-------------+---------------------------------------------------------------------------------------------------
observed |
e.v1 | 25.73631
e.v2 | 0 14.38707
e.v3 | 0 0 35.64868
e.v4 | 0 0 0 2.837749
e.v5 | 0 0 0 0 6.702231
e.v6 | 0 0 0 0 0 19.89547
e.v7 | 0 0 0 0 0 0 317.6616
e.v8 | 0 0 0 0 0 0 0 160.4738
e.v9 | 0 0 0 0 0 0 0 0 693.0864
-----------------------------------------------------------------------------------------------------------------
Intercepts of endogenous variables
| observed
alpha | v1 v2 v3 v4 v5 v6 v7 v8 v9
-------------+---------------------------------------------------------------------------------------------------
_cons | 29.57931 24.8 15.96552 9.951724 18.84828 17.28276 90.17931 109.7655 191.7793
-----------------------------------------------------------------------------------------------------------------
Covariances of exogenous variables
| latent
Phi | visual verbal speed
-------------+---------------------------------
latent |
visual | 21.73498
verbal | 7.338981 8.477162
speed | 38.11427 15.28808 244.0304
-----------------------------------------------
Means of exogenous variables
| latent
kappa | visual verbal speed
-------------+---------------------------------
mean | 0 0 0
-----------------------------------------------
When I am predicting the latent variables, I get the following results:
Code:
predict visual verbal speed, latent(visual verbal speed) sum visual verbal speed
PHP Code:
Variable | Obs Mean Std. Dev. Min Max
-------------+---------------------------------------------------------
visual | 145 -4.32e-09 4.000018 -9.592437 12.42246
verbal | 145 -1.23e-09 2.746986 -7.786823 7.084102
speed | 145 3.21e-08 13.91826 -40.7785 43.87231
My questions are:
1) Why the std. dev. of the predicted latent variables are smaller than the sqrt of the variances of the variables in matrix Phi?
2) If I use
Code:
estat framework, stand
Code:
egen vis = std(visual)
Emma
Comment