Validity for Latent Factors PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 JMC posted on Tuesday, March 19, 2013 - 4:44 pm
Hi Drs. Muthen,

Thanks in advance for your help. I am interested in including an index of factor validity, such as Coefficient H. I didn't see any mention of this in the manual. Can offer some guidance?

Thank you.
 Linda K. Muthen posted on Tuesday, March 19, 2013 - 5:02 pm
I am not familiar with Coefficient H. Mplus does not compute this as an option.
 JMC posted on Wednesday, March 20, 2013 - 6:26 pm
Thank you so much for getting back to me, this is very helpful. I organized my follow up questions by number for clarity.

1. Coefficient H is a analogous to a Cronbach's alpha for a latent factor. Is there a similar index that Mplus offers so I can report reliability for the latent factor?

2. I want to report my statistical assumptions (normality, linearity, and homoscedasticity) and I know these are available using Tech12 and Tech13, but I am not not using a mixture model. How can I access these values?

3. I would typically report Mahalanobis distance for outliers, but upon reading the manual I learned that this was only available for continuous DVs. I am using a count variable (total number correct) and a categorical variable (grades) for my DVs. What alternative to Mahalanobis would you recommend?

Thank you again for your time.
 Linda K. Muthen posted on Wednesday, March 20, 2013 - 6:37 pm
1. Mplus has no option for this coefficient.
2. You can use TYPE=MIXTURE. CLASSES = c (1);
3. There is no outlier detection for count variables.
 JMC posted on Sunday, March 31, 2013 - 6:30 am
Thank you so much, Dr. Muthen. I spent some time trying to work through this code, but haven't been able to run. The code that you gave me seems to be for a latent class analysis and I don't know how to make that work for my SEM. Can you explain further?

Also, one of my models has a count dependent variable and I cannot get it to converge. It works when I treat as continuous, but not as count. This is my code, can you offer any guidance?

TITLE: ITC SEM CATEGORICAL COUNT
DATA: File is C:\Users\cambria\Desktop\imputed_6.txt;

VARIABLE:
NAMES ARE ESL IEP ETH FARMS GENDER EFF1-EFF7 COG1-COG12
VAL1-VAL7 GRADES ITC;

USEVAR ARE ETH FARMS GENDER EFF1-EFF7 COG1-COG12
VAL1-VAL7 ITC;

CATEGORICAL IS ETH FARMS GENDER EFF1-EFF7 COG1-COG12 VAL1-VAL7;

COUNT IS ITC;

ANALYSIS:
ESTIMATOR = WLS;
ITERATIONS = 10000;

MODEL:
EFF BY EFF1-EFF7;
COG BY COG1-COG12;
VAL BY VAL1-VAL7;

ITC ON VAL (a1);
ITC ON EFF (a2);
ITC ON COG (a3);
ITC ON FARMS( a4);
ITC ON ETH (a5);
ITC ON GENDER (a8);
COG ON VAL (a6);
COG ON EFF (a7);


MODEL CONSTRAINT:
NEW (VALMED EFFMED MEDTOTAL);
VALMED = A6*A3;
EFFMED = A7*A3;
MEDTOTAL = VALMED + EFFMED;


OUTPUT:
SAMP STAND RES;


Thank you for your help!
 Linda K. Muthen posted on Sunday, March 31, 2013 - 2:44 pm
In the VARIABLE command add

CLASSES = c (1);

In the MODEL command before the BY statements add

%OVERALL%

In the ANALYSIS command add

TYPE=MIXTURE;

Regarding non-convergence, send the output and your license number to support@statmodel.com.
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: