ApolloChoiceModelling forum

Hello everyone,
I am trying to estimate a Mixed Logit model in Apollo. My MNL models run smoothly, but when I switch to Mixed Logit I get the following error:

Error in test && (as.character(e[[3]][[1]]) %in% c("apollo_mnl", "apollo_el", :
'length = 3' in coercion to 'logical(1)'.


I use the following code for analyse:

database <- choice_socio
# ============================================================
# 03_RPL_main_FIXED.R — Model only (no MWTP)
# Env attrs random (Normal); payments & interactions fixed; 2000 Sobol
# ============================================================

## 0) Clean + load Apollo + guard against any 'test' collisions
rm(list = setdiff(ls(), "choice_socio")); gc()
suppressPackageStartupMessages(library(apollo))
if ("test" %in% names(choice_socio)) choice_socio$test <- NULL
if (exists("test", envir = .GlobalEnv, inherits = FALSE)) rm("test", envir = .GlobalEnv)

## 1) Ensure choice (from choice1/2/3)
if (!("choice" %in% names(choice_socio))) {
stopifnot(all(c("choice1","choice2","choice3") %in% names(choice_socio)))
choice_socio$choice <- 1L*(choice_socio$choice1==1) +
2L*(choice_socio$choice2==1) +
3L*(choice_socio$choice3==1)
}

## 2) Apollo init + control
apollo_initialise()
database <- choice_socio
apollo_control <- list(
modelName = "RPL_effCont_interactions_moneyk_2000D",
modelDescr = "RPL: env random (Normal); payments & interactions fixed; 2000 Sobol",
indivID = "id",
panelData = TRUE,
mixing = TRUE,
nCores = 7
)

## 3) Draws (2000 Sobol)
apollo_draws <- list(
interDrawsType = "sobol",
interNDraws = 2000,
interNormDraws = c("d_wq_md","d_wq_lg","d_cul","d_fg","d_bio")
)

## 4) Parameters
apollo_beta <- c(
asc_sq = 0,
b_wq_md_mean = 0, b_wq_md_sd = 0.5,
b_wq_lg_mean = 0, b_wq_lg_sd = 0.5,
b_cul_mean = 0, b_cul_sd = 0.5,
b_fg_mean = 0, b_fg_sd = 0.5,
b_bio_mean = 0, b_bio_sd = 0.5,
b_time = 0, b_moneyk = 0,
b_only_time = 0, b_only_money = 0,
b_time_occ = 0,
b_moneyk_female = 0,
b_moneyk_inc_q2 = 0, b_moneyk_inc_q3 = 0, b_moneyk_inc_q4 = 0
)
apollo_fixed <- c()

## 5) Random coefficients (explicit named list + return)
apollo_randCoeff <- function(apollo_beta, apollo_inputs){
randCoeff <- list()
randCoeff[["b_wq_md"]] <- b_wq_md_mean + b_wq_md_sd * d_wq_md
randCoeff[["b_wq_lg"]] <- b_wq_lg_mean + b_wq_lg_sd * d_wq_lg
randCoeff[["b_cul"]] <- b_cul_mean + b_cul_sd * d_cul
randCoeff[["b_fg"]] <- b_fg_mean + b_fg_sd * d_fg
randCoeff[["b_bio"]] <- b_bio_mean + b_bio_sd * d_bio
return(randCoeff)
}

## 6) Likelihood (single-component pipeline; no list wrapper)
apollo_probabilities <- function(apollo_beta, apollo_inputs, functionality = "estimate"){
apollo_attach(apollo_beta, apollo_inputs); on.exit(apollo_detach(apollo_inputs), add=TRUE)
rcoefs <- apollo_inputs$apollo_randCoeff(apollo_beta, apollo_inputs)

# interactions per alt (wide)
time_occ_1 <- time1*occ_lake; time_occ_2 <- time2*occ_lake; time_occ_3 <- time3*occ_lake
money_f_1 <- money1_k*female; money_f_2 <- money2_k*female; money_f_3 <- money3_k*female
money_q2_1 <- money1_k*inc_q2; money_q2_2<- money2_k*inc_q2; money_q2_3<- money3_k*inc_q2
money_q3_1 <- money1_k*inc_q3; money_q3_2<- money2_k*inc_q3; money_q3_3<- money3_k*inc_q3
money_q4_1 <- money1_k*inc_q4; money_q4_2<- money2_k*inc_q4; money_q4_3<- money3_k*inc_q4

V <- list(
alt1 = rcoefs$b_wq_md*wq_md_eff1 + rcoefs$b_wq_lg*wq_lg_eff1 + rcoefs$b_cul*cul_imp_eff1 +
rcoefs$b_fg*fg_cont1 + rcoefs$b_bio*bio_cont1 +
b_time*time1 + b_moneyk*money1_k +
b_only_time*only_time1 + b_only_money*only_money1 +
b_time_occ*time_occ_1 + b_moneyk_female*money_f_1 +
b_moneyk_inc_q2*money_q2_1 + b_moneyk_inc_q3*money_q3_1 + b_moneyk_inc_q4*money_q4_1,
alt2 = rcoefs$b_wq_md*wq_md_eff2 + rcoefs$b_wq_lg*wq_lg_eff2 + rcoefs$b_cul*cul_imp_eff2 +
rcoefs$b_fg*fg_cont2 + rcoefs$b_bio*bio_cont2 +
b_time*time2 + b_moneyk*money2_k +
b_only_time*only_time2 + b_only_money*only_money2 +
b_time_occ*time_occ_2 + b_moneyk_female*money_f_2 +
b_moneyk_inc_q2*money_q2_2 + b_moneyk_inc_q3*money_q3_2 + b_moneyk_inc_q4*money_q4_2,
sq = asc_sq +
rcoefs$b_wq_md*wq_md_eff3 + rcoefs$b_wq_lg*wq_lg_eff3 + rcoefs$b_cul*cul_imp_eff3 +
rcoefs$b_fg*fg_cont3 + rcoefs$b_bio*bio_cont3 +
b_time*time3 + b_moneyk*money3_k +
b_only_time*only_time3 + b_only_money*only_money3 +
b_time_occ*time_occ_3 + b_moneyk_female*money_f_3 +
b_moneyk_inc_q2*money_q2_3 + b_moneyk_inc_q3*money_q3_3 + b_moneyk_inc_q4*money_q4_3
)

mnl_settings <- list(
alternatives = c(alt1=1, alt2=2, sq=3),
avail = c(alt1=1, alt2=1, sq=1),
choiceVar = choice,
V = V
)

# === single-component pipeline ===
P <- apollo_mnl(mnl_settings, functionality)
P <- apollo_panelProd(P, apollo_inputs, functionality)
if (isTRUE(apollo_inputs$apollo_control$mixing)) {
P <- apollo_avgInterDraws(P, apollo_inputs, functionality)
}
P <- apollo_prepareProb(P, apollo_inputs, functionality)
return(P)
}

## 7) Validate & estimate
apollo_inputs <- apollo_validateInputs()
model_rpl <- apollo_estimate(apollo_beta, apollo_fixed, apollo_probabilities, apollo_inputs)

## 8) Output (model only)
invisible(try(apollo_modelOutput(model_rpl), silent = TRUE))
invisible(try(apollo_saveOutput(model_rpl), silent = TRUE))


Thank you very much in advance.

Hi

1. which version of Apollo are you using?
2. the code is not standard, where did you get it from. E.g. you should not need this:

rcoefs <- apollo_inputs$apollo_randCoeff(apollo_beta, apollo_inputs)

Stephane

Hi Stephane

Thank you very much for the quick response.
I’m using Apollo version 0.3.5.
I realise now that my earlier script had a non-standard line. I was troubleshooting and mixed in pieces from older scripts.

It looks like the problem was related to stray objects and column creation in the global environment.

After I:
standardised money scaling (money1_k, money2_k, money3_k),
ensured only_time{1:3} and only_money{1:3} dummies exist, and
cleaned the global environment so no conflicting names remained,
the Mixed Logit ran fine.

For completeness, here’s a minimal working example I’m running now (wide data; environmental attributes random ~ Normal; payments fixed). If you notice anything non-standard, I’d appreciate any pointers. I suspect I may still have a few extra lines (e.g., some global-environment checks). I’m now extending the analysis to include interaction terms and increasing the number of draws.

Thank you very much for your kind guidance.
With best regards,
Nang

choice_socio <- choice_wide_with_socio
database <-choice_socio
# ---- Payment dummies: ensure columns exist (0 if missing) ----
for (nm in c(paste0("only_time",1:3), paste0("only_money",1:3))) {
if (!nm %in% names(database)) database[[nm]] <- 0L
}
# =========================
# Apollo Random Parameters Logit
# =========================
library(apollo)
apollo_initialise()

# ---- Load Data____
# database <-choice_wide_with_socio
stopifnot(exists("database"))

# ---- Reserve parameter names & purge collisions ----
param_names <- c(
"asc_sq","b_wq_md_mean","b_wq_md_sd","b_wq_lg_mean","b_wq_lg_sd",
"b_cul_mean","b_cul_sd","b_fg_mean","b_fg_sd","b_bio_mean","b_bio_sd",
"b_time","b_moneyk","b_only_money","b_only_time"
)
to_rm <- intersect(ls(envir = .GlobalEnv), param_names)
if (length(to_rm)) rm(list = to_rm, envir = .GlobalEnv)

# ---- Correct money scaling ----

mx <- max(database$money1, database$money2, database$money3, na.rm = TRUE)
for (i in 1:3) {
src <- paste0("money", i)
dst <- paste0("money", i, "k")
if (mx > 100) {
database[[dst]] <- database[[src]]/1000
} else {
database[[dst]] <- database[[src]]
}
}

# safe to check column name
if ("b_moneyk" %in% names(database)) stop("Rename column 'b_moneyk' in your data.")

database$choice <- as.integer(database$choice)
database$id <- as.integer(database$id)

###########################
# Drop rows with any NA in needed vars
vars_needed <- c("id","choice",
paste0("wq_md_eff",1:3), paste0("wq_lg_eff",1:3), paste0("cul_imp_eff",1:3),
paste0("fg_cont",1:3), paste0("bio_cont",1:3),
paste0("time",1:3), paste0("money",1:3,"k"),
paste0("only_time",1:3), paste0("only_money",1:3)
)

database <- database[complete.cases(database[, vars_needed]), ]
# ---- Control ----
out_dir <- file.path(getwd(), "apollo_out_RPL_moneyPer1k")
if (!dir.exists(out_dir)) dir.create(out_dir, recursive = TRUE)

apollo_control <- list(
modelName = "RPL_env_payments_Fixed_payment dummies",
modelDescr = "RPL with env attrs random (normal); time fixed; payment dummies",
indivID = "id",
panelData = TRUE,
mixing = TRUE,
nCores = 7,
outputDirectory = out_dir,
seed = 12345
)

# ---- Draws ----
apollo_draws <- list(
interDrawsType = "halton",
interNDraws = 1000,
interNormDraws = c("d_wq_md","d_wq_lg","d_cul","d_fg","d_bio")
)


# ---- Parameters ----
apollo_beta <- c(
asc_sq = 0,
b_wq_md_mean = 0, b_wq_md_sd = 0.1,
b_wq_lg_mean = 0, b_wq_lg_sd = 0.1,
b_cul_mean = 0, b_cul_sd = 0.1,
b_fg_mean = 0, b_fg_sd = 0.1,
b_bio_mean = 0, b_bio_sd = 0.1,
b_time = 0,
b_moneyk = 0,
b_only_time = 0, # <— new
b_only_money = 0 # <— new
)
apollo_fixed <- c()


# ---- Random coeffs ----
apollo_randCoeff <- function(apollo_beta, apollo_inputs){
b <- as.list(apollo_beta)
list(
b_wq_md = b$b_wq_md_mean + b$b_wq_md_sd * d_wq_md,
b_wq_lg = b$b_wq_lg_mean + b$b_wq_lg_sd * d_wq_lg,
b_cul = b$b_cul_mean + b$b_cul_sd * d_cul,
b_fg = b$b_fg_mean + b$b_fg_sd * d_fg,
b_bio = b$b_bio_mean + b$b_bio_sd * d_bio
)
}

# ---- Validate ----

# Load Apollo
if (!requireNamespace("apollo", quietly = TRUE)) install.packages("apollo")
library(apollo)
apollo_initialise()

if ("apollo_validateInputs" %in% ls()) rm(apollo_validateInputs)
if ("apollo" %in% ls()) rm(apollo)

apollo_inputs <- apollo::apollo_validateInputs()


# ---- Probabilities ----
apollo_probabilities <- function(apollo_beta, apollo_inputs, functionality="estimate"){
apollo_attach(apollo_beta, apollo_inputs); on.exit(apollo_detach(apollo_inputs), add=TRUE)

V <- list(
alt1 = b_wq_md*wq_md_eff1 + b_wq_lg*wq_lg_eff1 +
b_cul*cul_imp_eff1 + b_fg*fg_cont1 + b_bio*bio_cont1 +
b_time*time1 + b_moneyk*money1k +
b_only_time*only_time1 + b_only_money*only_money1,

alt2 = b_wq_md*wq_md_eff2 + b_wq_lg*wq_lg_eff2 +
b_cul*cul_imp_eff2 + b_fg*fg_cont2 + b_bio*bio_cont2 +
b_time*time2 + b_moneyk*money2k +
b_only_time*only_time2 + b_only_money*only_money2,

sq = asc_sq +
b_wq_md*wq_md_eff3 + b_wq_lg*wq_lg_eff3 +
b_cul*cul_imp_eff3 + b_fg*fg_cont3 + b_bio*bio_cont3 +
b_time*time3 + b_moneyk*money3k +
b_only_time*only_time3 + b_only_money*only_money3
)

mnl_settings <- list(
alternatives = c(alt1=1, alt2=2, sq=3),
avail = list(alt1=1, alt2=1, sq=1),
choiceVar = choice,
V = V
)

P <- list()
P[["model"]] <- apollo_mnl(mnl_settings, functionality)
P <- apollo_panelProd(P, apollo_inputs, functionality)
P <- apollo_avgInterDraws(P, apollo_inputs, functionality)
P <- apollo_prepareProb(P, apollo_inputs, functionality)
return(P)
}
# ---- Estimate & Output ----
model_rpl <- apollo_estimate(apollo_beta, apollo_fixed, apollo_probabilities, apollo_inputs)
apollo_modelOutput(model_rpl)
apollo_saveOutput(model_rpl)
##############################################################################

can you share the code and data with me outside the forum and I'll have a look

Hello dear Apollo community,

I am encountering the same issue the colleague had this last september. I am preparing an HCM with covariates and ordinal logit.

My model run well and has found convergence, but the following message appears when I use the function apollo_modelOutput(model):

"Model run by t.feldens using Apollo 0.3.7 on R 4.5.3 for Windows.
Please acknowledge the use of Apollo by citing Hess & Palma (2019)
DOI 10.1016/j.jocm.2019.100170
www.ApolloChoiceModelling.com

Model name : HCM3
Model description : Hybrid choice model using ordered measurement model for indicators of risk (Q's) and covariates
Model run at : 2026-04-15 09:40:18.569851
Estimation method : bgw
Estimation diagnosis : Relative function convergence (favorable)
Error in is.null(model$hessianMethodsAttempted) || model$hessianMethodsAttempted == :
'length = 2' in coercion to 'logical(1)''".



Here is the complete script I am using:



library(apollo)
library(dplyr)
library(haven)


apollo_initialise()

### Set core controls
apollo_control = list(
modelDescr = "Hybrid choice model using ordered measurement model for indicators of risk and demographics",
modelName = "HCM3",
indivID = "attribute_5",
nCores= 45,
outputDirectory = "output"
)



choiceAnalysis_settings = list(
alternatives = c(visita =1, test=2, optout=3),
choiceVar = database$Choice,
explanators = database[,c("Q20_1","Q20_2","Q20_3", "Q20_4", "Q20_5", "Q20_6", "Q20_7", "Q20_8")]
)

apollo_choiceAnalysis(choiceAnalysis_settings, apollo_control, database)


apollo_beta <- c(
asc_visita = 0,
asc_test = 0,
asc_optout = 0,
b_visita_specificita = 0,
b_visita_prezzo = 0,
b_visita_luogo = 0,
b_test_specificita = 0,
b_test_prezzo = 0,
b_test_luogo = 0,
lambda = 1,
b_cleta_50_59_visita = 0,
b_cleta_50_59_test = 0,
b_cleta_60_69_visita = 0,
b_cleta_60_69_test = 0,
b_cleta_70_visita = 0,
b_cleta_70_test = 0,
b_NordOv_test = 0,
b_NordOv_visita = 0,
b_NordEs_test = 0,
b_NordEs_visita = 0,
b_Centro_test = 0,
b_Centro_visita = 0,
b_Nopartner_test = 0,
b_Nopartner_visita = 0,
b_Educ_med_elem_test = 0,
b_Educ_med_elem_visita = 0,
b_Educ_laurea_test = 0,
b_Educ_laurea_visita = 0,
gamma_q20_1 = 1,
gamma_q20_2 = 0,
gamma_q20_3 = 0,
gamma_q20_4 = 0,
gamma_q20_5 = 0,
gamma_q20_6 = 0,
gamma_q20_7 = 0,
gamma_q20_8 = 0,
gamma_eta_over_50 = 0,
gamma_studio_laurea = 0,
tau_q20_1_1 = -5,
tau_q20_1_2 = -4,
tau_q20_1_3 = -3,
tau_q20_1_4 = -2,
tau_q20_1_5 = -1,
tau_q20_1_6 = 1,
tau_q20_1_7 = 2,
tau_q20_1_8 = 3,
tau_q20_1_9 = 4,
tau_q20_2_1 = -5,
tau_q20_2_2 = -4,
tau_q20_2_3 = -3,
tau_q20_2_4 = -2,
tau_q20_2_5 = -1,
tau_q20_2_6 = 1,
tau_q20_2_7 = 2,
tau_q20_2_8 = 3,
tau_q20_2_9 = 4,
tau_q20_3_1 = -5,
tau_q20_3_2 = -4,
tau_q20_3_3 = -3,
tau_q20_3_4 = -2,
tau_q20_3_5 = -1,
tau_q20_3_6 = 1,
tau_q20_3_7 = 2,
tau_q20_3_8 = 3,
tau_q20_3_9 = 4,
tau_q20_4_1 = -5,
tau_q20_4_2 = -4,
tau_q20_4_3 = -3,
tau_q20_4_4 = -2,
tau_q20_4_5 = -1,
tau_q20_4_6 = 1,
tau_q20_4_7 = 2,
tau_q20_4_8 = 3,
tau_q20_4_9 = 4,
tau_q20_5_1 = -5,
tau_q20_5_2 = -4,
tau_q20_5_3 = -3,
tau_q20_5_4 = -2,
tau_q20_5_5 = -1,
tau_q20_5_6 = 1,
tau_q20_5_7 = 2,
tau_q20_5_8 = 3,
tau_q20_5_9 = 4,
tau_q20_6_1 = -5,
tau_q20_6_2 = -4,
tau_q20_6_3 = -3,
tau_q20_6_4 = -2,
tau_q20_6_5 = -1,
tau_q20_6_6 = 1,
tau_q20_6_7 = 2,
tau_q20_6_8 = 3,
tau_q20_6_9 = 4,
tau_q20_7_1 = -5,
tau_q20_7_2 = -4,
tau_q20_7_3 = -3,
tau_q20_7_4 = -2,
tau_q20_7_5 = -1,
tau_q20_7_6 = 1,
tau_q20_7_7 = 2,
tau_q20_7_8 = 3,
tau_q20_7_9 = 4,
tau_q20_8_1 = -5,
tau_q20_8_2 = -4,
tau_q20_8_3 = -3,
tau_q20_8_4 = -2,
tau_q20_8_5 = -1,
tau_q20_8_6 = 1,
tau_q20_8_7 = 2,
tau_q20_8_8 = 3,
tau_q20_8_9 = 4)

apollo_fixed <- c("asc_optout", "gamma_q20_1")

apollo_draws = list(
interDrawsType="halton",
interNDraws=50, #for test, later on I will increase it
interUnifDraws=c(),
interNormDraws=c("chi"),

intraDrawsType="",
intraNDraws=0,
intraUnifDraws=c(),
intraNormDraws=c()
)

apollo_randCoeff=function(apollo_beta, apollo_inputs){
randcoeff = list()
randcoeff[["LV"]] = chi + gamma_eta_over_50*cleta_over_50 + gamma_studio_laurea*Educ_laurea
return(randcoeff)
}


apollo_inputs <- apollo_validateInputs()


apollo_probabilities <- function(apollo_beta, apollo_inputs, functionality = "estimate"){

apollo_attach(apollo_beta, apollo_inputs)
on.exit(apollo_detach(apollo_beta, apollo_inputs))

P <- list()
V <- list()


### 0 - Indicators OL
ol_settings1 = list(outcomeOrdered = Q20_1,
V = gamma_q20_1*LV,
tau = list(tau_q20_1_1, tau_q20_1_2, tau_q20_1_3, tau_q20_1_4, tau_q20_1_5, tau_q20_1_6, tau_q20_1_7,
tau_q20_1_8, tau_q20_1_9),
rows = (Question_number=="task1a"))
P[["mQ20_1"]] = apollo_ol(ol_settings1, functionality)


ol_settings2 = list(outcomeOrdered = Q20_2,
V = gamma_q20_2*LV,
tau = list(tau_q20_2_1, tau_q20_2_2, tau_q20_2_3, tau_q20_2_4, tau_q20_2_5, tau_q20_2_6, tau_q20_2_7,
tau_q20_2_8, tau_q20_2_9),
rows = (Question_number=="task1a"))
P[["mQ20_2"]] = apollo_ol(ol_settings2, functionality)


ol_settings3 = list(outcomeOrdered = Q20_3,
V = gamma_q20_3*LV,
tau = list(tau_q20_3_1, tau_q20_3_2, tau_q20_3_3, tau_q20_3_4, tau_q20_3_5, tau_q20_3_6, tau_q20_3_7,
tau_q20_3_8, tau_q20_3_9),
rows = (Question_number=="task1a"))
P[["mQ20_3"]] = apollo_ol(ol_settings3, functionality)


ol_settings4 = list(outcomeOrdered = Q20_4,
V = gamma_q20_4*LV,
tau = list(tau_q20_4_1, tau_q20_4_2, tau_q20_4_3, tau_q20_4_4, tau_q20_4_5, tau_q20_4_6, tau_q20_4_7,
tau_q20_4_8, tau_q20_4_9),
rows = (Question_number=="task1a"))
P[["mQ20_4"]] = apollo_ol(ol_settings4, functionality)


ol_settings5 = list(outcomeOrdered = Q20_5,
V = gamma_q20_5*LV,
tau = list(tau_q20_5_1, tau_q20_5_2, tau_q20_5_3, tau_q20_5_4, tau_q20_5_5, tau_q20_5_6, tau_q20_5_7,
tau_q20_5_8, tau_q20_5_9),
rows = (Question_number=="task1a"))
P[["mQ20_5"]] = apollo_ol(ol_settings5, functionality)

ol_settings6 = list(outcomeOrdered = Q20_6,
V = gamma_q20_6*LV,
tau = list(tau_q20_6_1, tau_q20_6_2, tau_q20_6_3, tau_q20_6_4, tau_q20_6_5, tau_q20_6_6, tau_q20_6_7,
tau_q20_6_8, tau_q20_6_9),
rows = (Question_number=="task1a"))
P[["mQ20_6"]] = apollo_ol(ol_settings6, functionality)

ol_settings7 = list(outcomeOrdered = Q20_7,
V = gamma_q20_7*LV,
tau = list(tau_q20_7_1, tau_q20_7_2, tau_q20_7_3, tau_q20_7_4, tau_q20_7_5, tau_q20_7_6, tau_q20_7_7,
tau_q20_7_8, tau_q20_7_9),
rows = (Question_number=="task1a"))
P[["mQ20_7"]] = apollo_ol(ol_settings7, functionality)

ol_settings8 = list(outcomeOrdered = Q20_8,
V = gamma_q20_8*LV,
tau = list(tau_q20_8_1, tau_q20_8_2, tau_q20_8_3, tau_q20_8_4, tau_q20_8_5, tau_q20_8_6, tau_q20_8_7,
tau_q20_8_8, tau_q20_8_9),
rows = (Question_number=="task1a"))
P[["mQ20_8"]] = apollo_ol(ol_settings8, functionality)



### 1. Utilities
V[['visita']] = asc_visita +
b_visita_specificita*visita_specificita +
b_visita_prezzo*visita_prezzo +
b_visita_luogo*visita_luogo + lambda*LV +
b_cleta_50_59_visita*cleta50_59 + b_cleta_60_69_visita*cleta60_69 + b_cleta_70_visita*cleta70 +
b_NordOv_visita*NordOv + b_NordEs_visita*NorEs + b_Centro_visita*Centro +
b_Nopartner_visita*Nopartner +
b_Educ_med_elem_visita*Educ_med_elem + b_Educ_laurea_visita*Educ_laurea


V[['test']] = asc_test +
b_test_specificita*test_specificita +
b_test_prezzo*test_prezzo +
b_test_luogo*test_luogo + lambda*LV +
b_cleta_50_59_test*cleta50_59 + b_cleta_60_69_test*cleta60_69 + b_cleta_70_test*cleta70 +
b_NordOv_test*NordOv + b_NordEs_test*NorEs + b_Centro_test*Centro +
b_Nopartner_test*Nopartner +
b_Educ_med_elem_test*Educ_med_elem + b_Educ_laurea_test*Educ_laurea

V[['optout']] = asc_optout

### 2. MNL
mnl_settings <- list(
alternatives = c(visita = 1, test = 2, optout = 3),
choiceVar = Choice,
V = V
)

### 3. Prob per obs
# P[["choice"]] = apollo_mnl(mnl_settings, functionality)
P[["choice"]] = apollo_mnl(mnl_settings, functionality)

P[["model"]] = P[["choice"]] * P[["mQ20_1"]] * P[["mQ20_2"]] * P[["mQ20_3"]] * P[["mQ20_4"]] * P[["mQ20_5"]] * P[["mQ20_6"]] * P[["mQ20_7"]] * P[["mQ20_8"]]

### 4. Panel structure considered
P = apollo_panelProd(P, apollo_inputs, functionality)

### 5.Draws (absorting chi)
P = apollo_avgInterDraws(P, apollo_inputs, functionality)

### 6. Preparing obs
P = apollo_prepareProb(P, apollo_inputs, functionality)
return(P)
}

model = apollo_estimate(apollo_beta, apollo_fixed, apollo_probabilities, apollo_inputs)

apollo_modelOutput(model)



Thank you so much for any help possible,

Hi

are you able to share your code and data with me and I can look into it

Stephane