Lagrange Multiplier Test

1 reply [Last post]
brauer's picture
Offline
Joined: 01/28/2012

Hi, is there a way to get the Lagrange Multiplier Test with OpenMx? I am usually not a fan of this type of data-driven test but I am currently analyzing a dataset where a LM Test might help me understand what is going on. I am running a four-factor CFA model with three indicators. The fit is OK but not great. The correlation residuals tell me that I am drastically underestimating the relationship between one indicator of the first factor and one indicator of the second factor. But then, when I allow the measurement errors of these indicators to covary, the model fit hardly improves and the estimated covariance between the two measurement errors is close to zero. I don't get it. Maybe a LM Test will help me understand what is going on in these data.

mhunter's picture
Offline
Joined: 07/31/2009
I'm not currently that

I'm not currently that familiar with Lagrange Multipliers or their tests, but I'm fairly certain that OpenMx does not offer or use them right now. The optimizer (NPSOL) uses Lagrange Multipliers to check and operate with constraints, but that is slightly different from what I think you are looking for.

My understanding is that you're looking for some kind of modification index. These have been brought up before, and I think someone came up with some add-ons to OpenMx that can do them

Here's the thread:
http://openmx.psyc.virginia.edu/thread/1019

Hope this helps get you unstuck!