Testing Difference in Parameters PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Chelsea Jin posted on Monday, March 04, 2013 - 2:57 pm
Dear Drs. Muthen,

I have a question regarding how to test the difference between parameters.

Let's say - I have a equation, e.g.,

Y on X1 X2 X3;

The parameters of X1, X2 & X3 can be hence obtained - suppose that they are beta1 for X1, beta2 for X2, and beta3 for X3.

My purpose is to test the difference between beta1 & beta2 (beta2-beta1), beta1 & beta3 (beta3-beta1), and beta2 & beta3(beta3-beta2). My question is how to write codes in Mplus to obtain these differences.

I'd appreciate you would reply.
 Linda K. Muthen posted on Monday, March 04, 2013 - 3:45 pm
MODEL:

y on x1 (beta1)
x2 (beta2)
x3 (beta3);

MODEL CONSTRAINT:
NEW (diff);

diff = beta1 - beta2;

etc.
 Chelsea Jin posted on Tuesday, March 05, 2013 - 7:41 am
Thanks for your reply, Dr. Muthen. However, an error message occurred as

*** ERROR in MODEL command
Unknown variable(s) in an ON statement: (BETA1)

Do I have to define beta1, beta2, etc.? And how? I'd appreciate your reply.
 Linda K. Muthen posted on Tuesday, March 05, 2013 - 7:48 am
Please send your output and license number to support@statmodel.com.
 Maria Pavlova posted on Friday, October 20, 2017 - 2:30 am
Hi,
is there any simple way to test a difference between STANDARDIZED coefficients from two different equations (same predictors, different outcomes) when both predictor and outcome variables are latent? I test the difference between model parameters using model constraint. However, it enables testing differences between unstandardized parameters only. I know that I cannot fix the variance of the DV to 1 (only the residual variance, which doesn't serve the purpose). I tried using standardized manifest indicators for all latent variables, but to my surprise, the unstandardized output still shows vastly different variances of latent variables even for predictors. Any advice will be much appreciated!
 Bengt O. Muthen posted on Friday, October 20, 2017 - 5:35 pm
You have to express the standardized effect in Model Constraint, so using Model parameter labels that can be used to express the SDs used in the standardization as well.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: