LTA Analysis

[Kyle Rehn]

Is it possible to conduct latent transition analysis using the lavaan package? Thanks so much!

Kyle

[Mauricio Garnier-Villarreal]

No, for such model I think the best option within R is depmixS4 or LatentGold (proprietary)

[Mauricio Garnier-Villarreal]

I have tutorial on it here

https://github.com/maugavilla/well_hello_stats/blob/main/tutorials/12_HMM_depmixS4_cat.md

[Kyle Rehn]

Hey! This was SUPER helpful and I really appreciate your help. I think I got most of the code figured out finally except for the covariates part. I want to include Wave as the time interval period over time and Sex as the covariate. The code works but I can't see how Sex plays a part in the output (cant see it anywhere). Do you have any recommendations? Here is my code below. 

ST <- depmix(
  response = list(Physical ~ 1, Counseling ~ 1, NeedCare ~ 1, SuicideThink ~ 1),
  data = HMM,
  nstates = 4,
  transition = ~ Wave,
  ntimes = as.numeric(table(HMM$AID)),
  family = list(
    multinomial("identity"),
    multinomial("identity"),
    multinomial("identity"),
    multinomial("identity"),
    multinomial("identity"),
    multinomial("identity"),
    multinomial("identity"),
    multinomial("identity")
  ),
  covariates = ~ Sex
)

Thanks so much!

Kyle

[Mauricio Garnier-Villarreal]

Kyle

As this is no longer a lavaan question please email me directly, and I can help you on that

(I should add a covariate section to the titorial then jaja)