| Message/Author |
|
|
|
|
| I ran a negative binomial model to predict an ordered outcome with 6 levels and a large (2/3 of sample) number of zero values. Measured variable values ranged from 0 through 5. The output reported that a value of the criterion variable had to be fixed at its extreme, which was 15, but also that the model terminated normally. Three questions: what does this mean? Is there something to fix in the data and how? And, can I trust the output? Thank you. |
|
|
|
|
| You say "negative binomial" and also "an ordered outcome". They wouldn't be one and the same because negbin is for count outcomes. Perhaps you mean that the count variable predicts an ordinal outcome but that isn't clear. You may want to send your output to Support along with your license number. |
|
|
| Adam Garber posted on Tuesday, July 23, 2019 - 12:57 am
|
|
|
Hello,
1. In Mplus is it possible to specify a poisson gamma regression for a highly positive skewed continuous outcome?
2. In Mplus is it possible to specify a gamma regression?
3. What modeling approach would you recommend for a highly positive skewed continuous outcome with true zeros?
I have read that the negative binomial model can be used for non-integer values. I am interested in comparing fit between the censored, the poisson gamma, and gamma regression model approaches. |
|
|
|
|
Why don't you try the Mplus skew-normal and skew-t options. See the paper on our website:
Asparouhov, T. & Muthén B. (2015). Structural equation models and mixture models with continuous non-normal skewed distributions. Structural Equation Modeling: A Multidisciplinary Journal, DOI:10.1080/10705511.2014.947375. (Download Mplus inputs and outputs). |
|
|
| Adam Garber posted on Wednesday, July 24, 2019 - 9:25 am
|
|
|
Thanks for the quick response.
I ran the SKEWT general SEM model and received a number of error messages that I don't know how to interpret/resolve:
Y2 has skew = 1.4
Y1 has skew = 0.84
-----
Analysis:
TYPE = general;
estimator = MLR;
coverage = 0;
DISTRIBUTION = SKEWT;
processors = 8;
Model:
M on X;
Y1 on X M W X_W M_W;
Y2 on X M W X_W M_W;
Y1 with Y2;
---
WARNING: DUE TO A LOW DF ESTIMATE IN CLASS 1 THE ESTIMATED SKEWNESS IN THAT CLASS IS INFINITY.
THE DISTRIBUTIONAL ASSUMPTIONS OF THE SKEW-T DISTRIBUTION MAY NOT BE APPROPRIATE.
ONE OR MORE PARAMETERS WERE FIXED TO AVOID SINGULARITY OF THE
INFORMATION MATRIX. THE SINGULARITY IS MOST LIKELY BECAUSE THE
MODEL IS NOT IDENTIFIED, OR BECAUSE OF EMPTY CELLS IN THE JOINT
DISTRIBUTION OF THE CATEGORICAL VARIABLES IN THE MODEL.
THE FOLLOWING PARAMETERS WERE FIXED:
Parameter 17, Y_TCDEP WITH Y_TPDEP
Parameter 18, Y_TCDEP
THIS MAY ALSO BE DUE TO RESIDUAL VARIANCES CONVERGING TO 0.
THESE RESIDUAL VARIANCES AND CORRESPONDING COVARIANCES ARE FIXED TO 0. |
|
|
|
|
Perhaps your outcomes have strong floor or ceiling effects in which case Skew-t and Skew-normal are not suitable.
If this doesn't help, send your full output and data to Support along with your license number. |
|
|
| Adam Garber posted on Wednesday, July 24, 2019 - 6:50 pm
|
|
|
| Thanks this is very helpful and appreciated. |
|
|
| Back to top |
