Hi all,
I'm trying to impute factors after testing my model for CMV and therefore included a common latent factor. Unfortunately, I always get the same error: "A sample of parameter values was inadmissible". I've checked some earlier suggested troubleshooting options - the model runs fine with and without CLF, sufficient sample size (n=304), no missing data, no spaces in variable names, no negative error variance but when I check the standardized residual covariance matrix, I find loads of negative values. Can this cause the error? And if so, do you may have a suggestion on how to fix it? I've tried to constraint the latent factors' variance to 0,005. This led to a substantial decrease of all loadings AND I still have negative covariances.
I'd appreciate any hint!
Thanks!
Thank you Cristian Ramos-Vera. Any thoughts on how to solve the problem? I ran out of ideas.
It is possible to review the covariance error re-specification and correlate those items, but I guess the complex model is similar to a bifactor model, I don't know if it is possible to do that in AMOS.
Thanks Cristian Ramos-Vera, looked at them but found some sources that said whenever possible it should be avoided to correlate error items. Since it seems to be a discussed topic, I hesitated and tried to find another solution. For anyone who might stumble across this thread and wonders how I solved it in the end - I went way back and rechecked for normally distributed data. Found some variables with really high kurtosis, removed them - redid EFA and CFA and it finally worked :) Also found this thread to be insightful: https://www.ibm.com/support/pages/amos-states-some-model-variance-values-are-negative. Hope it helps anyone!
Hallo Rosa Cosima Spitzer,
ich bin gerade auf das gleiche Problem gestoßen und bei der Suche nach einer Lösung über diesen Thread gestolpert.
Wie bist du bei der Suche nach normalverteilten Daten vorgegangen und wie konntest du diese hohe Kurosis identifizieren?
Ich würde mich sehr über eine Antwort freuen!
(leider kann ich den Link nicht öffnen)
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running