Re: ordered categorical variables

[Jeremy Miles]

If you're running the same model on each year's data, then presumably it's something about the data in the years where it doesn't converge.

Without more information, I'm not sure it's possible to determine the problem.

(Your model code, or data, would make it easier to answer).

Is there a reason you're not fitting a longitudinal model?

Jeremy

On Mon, 24 Mar 2025 at 13:31, Sara <sarahfi...@gmail.com> wrote:

 I’m working on a dataset with questions asked to the same subjects every 2 years from 2000 to 2006. These questions relate to emotional state and have 4 possible responses: never, most of the time, all of the time, and always—making them ordinal categorical variables. I need to perform a confirmatory analysis, starting with 2 factors: positive and negative emotions.   Problem: The model fails to converge in most years, e.g., 2010.  If anyone has suggestions, I’d really appreciate. 

> fit_two_2010 <- cfa(model_2010, data = data2010, ordered = names(data2010), std.lv = TRUE) Warning message: lavaan->lav_lavaan_step11_estoptim(): Model estimation FAILED! Returning starting values.

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/lavaan/3b644cbe-746a-4982-ba29-06d799690248n%40googlegroups.com.

[Yves Rosseel]

> negative emotions. / *Problem:* The model *fails to converge in most
> years*, e.g., 2010. *If anyone has suggestions, I’d really appreciate.

As a general approach to handle non-convergence issues:

1) remove the std.lv = TRUE argument
2) use rstarts = 20L (or higher) to try again 20 more times with random
starting values
3) use bounds = TRUE to try again with bounded estimation

Yves.