|
|
 |
| Growth Model with Count Outcome |
 
|
| Message/Author |
|
|
|
|
Hello,
I'm trying to run a latent growth model with a count outcome variable across 6 time points. I have a baseline predictor (trait aggression, continuous) and several baseline time-invariant covariates (count, categorical, continuous).
1) When I run the unconditional model, the mean intercept is negative. Given that I can't have a negative start value with a count DV, I've read that it is appropriate to fix your intercept mean at 0, using [i@0]. Is this correct? If not, how can I specify the model to only allow integers for the mean intercept?
2) If when I run the conditional model with all covariates, and my main predictor (Trait aggression, continuous) has a positive slope growth factor predicting an overall negative slope of DV (offending, count variable), is it correct to say that this means that those with higher levels of trait aggression (IV) have a *slower* decline in offending over time? In other words, those with low levels of trait aggression (IV) have a faster decline in offending (DV) over time?
Thank you,
Emily Robertson |
|
|
|
|
1) See my answer to Mary M Mitchell this afternoon.
2) Yes, this is the way to interpret it. |
|
|
|
|
Thanks for your assistance.
Can you confirm if the interpretation for question #2 is the same if you have a categorical predictor variable and a continuous outcome?
Thanks again. |
|
|
|
|
I have an additional question, and I apologize for the double post. When I don't constrain the mean intercept to be 0 in my growth model, which is what I concluded based on the post you directed me to, I get the following two errors:
THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES MAY NOT BE
TRUSTWORTHY FOR SOME PARAMETERS DUE TO A NON-POSITIVE DEFINITE
FIRST-ORDER DERIVATIVE PRODUCT MATRIX. THIS MAY BE DUE TO THE STARTING
VALUES BUT MAY ALSO BE AN INDICATION OF MODEL NONIDENTIFICATION. THE
CONDITION NUMBER IS 0.843D-18. PROBLEM INVOLVING THE FOLLOWING PARAMETER:
Parameter 49, WHITE
ONE OR MORE PARAMETERS WERE FIXED TO AVOID SINGULARITY OF THE
INFORMATION MATRIX. THE SINGULARITY IS MOST LIKELY DUE TO THE
MODEL IS NOT IDENTIFIED, OR DUE TO A LARGE OR A SMALL PARAMETER
ON THE LOGIT SCALE. THE FOLLOWING PARAMETERS WERE FIXED:
Parameter 4, S4SROV
The first error, I've seen on other posts is OK if your outcome variable is a count variable. However, I just want to confirm whether the second error is also OK or if this does in fact mean model non-identification? |
|
|
|
|
7:53 question: Yes.
8:58 question: The second message indicates model non-identification. |
|
|
| Back to top |
|
|
