Error using externVar: "classmb is missing

[HannaMuetze]

Dear all, I'm investigating the relationship between latent classes and external predictors using the code and datastructure below. Does anybody can help me? Thanks so much! Kind regards

XextHess <- externVar(TL_MOD_FMMAccmod6_3, # primary model
                       classmb = ~gold_msi, # external predictor variable 
                        subject = "id", # subject identifier
                       data = TL_PRE_Long4, # data
                      method = "twoStageJoint") 

Error message: "Fehler in .Contlcmm(fixed = fixed, mixture = mixture, random = random, : Argument "classmb" fehlt (ohne Standardwert)" --> english: "argument classmb is missing".

This is my data structure: id time accuracy gold_msi 1 0 0.9102564 80 2 0 0.8974359 58 3 0 0.8974359 57 4 0 0.8076923 43 5 0 0.8846154 69 6 0 0.1666667 70

[VivianePhilipps]

Hi,

could you give me information about the primary model, please? I need to run a similar model (same function, same number of variables, etc) so that I can check where the error occurs. Or do you have a reproducible example using the data provided in the package (data_lcmm, paquid)?

Viviane

[HannaMuetze]

Hi, thanks for your reply. I attached the dataset. You can find the complete code below. Thank you so much for your help! Kind regards, Hanna

library(readxl)
TL_PRE_Long4 <- read_excel("Data/Objects/TL_PRE_Long4_subset.xlsx")

# 1-class model for "minit"
FMMAccmod6_1 <- lcmm(accuracy ~ poly(time, degree = 2, raw = TRUE),
                     random =~ 1 + poly(time, degree = 2, raw = TRUE), 
                     ng = 1, 
                     subject = "id", data = TL_PRE_Long4,
                     link = "9-quant-splines")
# 3-class model (=primary model)
TL_MOD_FMMAccmod6_3  <- lcmm:: gridsearch(rep = 50, maxiter = 100, minit = FMMAccmod6_1, 
                                   m = lcmm(accuracy ~ poly(time, degree = 2, raw = TRUE),
                                            random = ~ 1 + poly(time, degree = 2, raw = TRUE),
                                            subject = "id", data = TL_PRE_Long4,
                                            link = "9-quant-splines", 
                                            mixture = ~ 1 + poly(time, degree = 2, raw = TRUE),
                                            ng = 3,
                                            nwg = F))
XextHess <- externVar(TL_MOD_FMMAccmod6_3, # primary model
                       classmb = ~gold_msi, # external predictor variable 
                        subject = "id", # subject identifier
                       data = TL_PRE_Long4, # data
                      method = "twoStageJoint") 

TL_PRE_Long4_subset.xlsx

[CecileProust-Lima]

Hi, the code of externVar was changed on the github version. It would be worth trying with it, and trying also with the "conditional" technique. Cécile