Code for Latent class model with covariates
Posted: 06 Nov 2023, 09:52
Dear Stephane,
I´ve tried to write a code for a latent class model with covariates.
When defining the utility I used the following coding:
V[["class_a"]] = delta_a + gamma_EinstFin_7_a*EinstFin_7 + gamma_FinLue_Angehoerigeetc_a*FinLue
V[["class_b"]] = delta_b + gamma_EinstFin_7_b*EinstFin_7 + gamma_FinLue_Angehoerigeetc_b*FinLue
EinstFin_7 is a metric variable
FinLue is a variable that is separated into two categories, one of this categories is FinLue_Angehoerigeetc (see transformations of this variable below:
Data_Covariates$FinLue_kat <- NA
Data_Covariates$FinLue_kat[Data_Covariates$FinLue %in% c(1)] <- "HzP"
Data_Covariates$FinLue_kat[Data_Covariates$FinLue %in% c(2,3)] <- "Angehoerigeetc"
Data_Covariates$FinLue_Angehoerigeetc <- ifelse(Data_Covariates$FinLue_kat=="Angehoerigeetc",1,0)
I am unsure if it is the right approach to multiply gamma_FinLue_Angehoerigeetc_a with FinLue (so with the whole variable) or if gamma_FinLue_Angehoerigeetc_a needs to be multiplied with FinLue_Angehoerigeetc
In my ML with covariates I did specify the coefficient like this:
b_HoeheEEE = b_HoeheEEE_ALLE + b_HoeheEEE_ FinLue_Angehoerigeetc * FinLue_Angehoerigeetc
That´s why I am a bit confused.
Any advice for clarification is highly appreciated.
Thanks and best,
Julia
I´ve tried to write a code for a latent class model with covariates.
When defining the utility I used the following coding:
V[["class_a"]] = delta_a + gamma_EinstFin_7_a*EinstFin_7 + gamma_FinLue_Angehoerigeetc_a*FinLue
V[["class_b"]] = delta_b + gamma_EinstFin_7_b*EinstFin_7 + gamma_FinLue_Angehoerigeetc_b*FinLue
EinstFin_7 is a metric variable
FinLue is a variable that is separated into two categories, one of this categories is FinLue_Angehoerigeetc (see transformations of this variable below:
Data_Covariates$FinLue_kat <- NA
Data_Covariates$FinLue_kat[Data_Covariates$FinLue %in% c(1)] <- "HzP"
Data_Covariates$FinLue_kat[Data_Covariates$FinLue %in% c(2,3)] <- "Angehoerigeetc"
Data_Covariates$FinLue_Angehoerigeetc <- ifelse(Data_Covariates$FinLue_kat=="Angehoerigeetc",1,0)
I am unsure if it is the right approach to multiply gamma_FinLue_Angehoerigeetc_a with FinLue (so with the whole variable) or if gamma_FinLue_Angehoerigeetc_a needs to be multiplied with FinLue_Angehoerigeetc
In my ML with covariates I did specify the coefficient like this:
b_HoeheEEE = b_HoeheEEE_ALLE + b_HoeheEEE_ FinLue_Angehoerigeetc * FinLue_Angehoerigeetc
That´s why I am a bit confused.
Any advice for clarification is highly appreciated.
Thanks and best,
Julia