 Score of latent response variable    Message/Author  gibbon lab posted on Monday, January 23, 2012 - 1:52 pm
For a latent response variable underlying one categorical variable, thresholds of the latent response variable are provided by Mplus in the output. How to get the value (or score) for this kind of latent variable?
Here is a simple example code:
categorical are x;
model:
x on z;
Where z is a exogenous variable. How to get the value (or score) of the latent response variable underlying x? Thanks.  gibbon lab posted on Wednesday, January 25, 2012 - 12:35 pm
Hi Professor Muthen,

Maybe I did not make my question clear in the previous post. Let me give a simple example. In a probit model,let y be the binary (0/1) outcome and x the scalar predictor. The probit model is
ystar=b0+b1*x+epsilon,
where b0 and b1 are intercept and coefficient respectively, epsilon follows a standard normal distribution and y=1{ystar>0}. Here 1{.} is an indicator function. Is there a command in Mplus to calculate ystar? Thanks a lot.  Linda K. Muthen posted on Wednesday, January 25, 2012 - 1:39 pm
ANALYSIS: PARAMETERIZATION = Theta;
MODEL:
f BY u1; f@1; u1@.00001;
f ON x1 x3;  gibbon lab posted on Monday, January 30, 2012 - 12:35 pm
Hi Linda,

Sorry, I do not quite understand your coding.
Is u1 here the categorical variable?
Why is it necessary to set the loading of u1 to be almost zero? Thanks a lot.

ANALYSIS: PARAMETERIZATION = Theta;
MODEL:
f BY u1; f@1; u1@.00001;
f ON x1 x3;  Linda K. Muthen posted on Monday, January 30, 2012 - 12:42 pm
Yes, u1 is the categorical variable. It is the residual variance not the loading that is set close to zero.    Topics | Tree View | Search | Help/Instructions | Program Credits Administration