I am using a multivariate joint model to test for main and interactive effects of two longitudinal timeseries on survival. It’s working almost as expected, except that one of the two longitudinal effects yields weirdly inflated associations. I have tried a univariate joint model, various association structures, and also working with different data subsets, but the effect remains, or becomes even bigger (like, 10^9 ridiculously big).

I don’t know what this means and how I can fix it. Because I am new to bayesian modelling, I am also not sure which additional information provide.

EDIT: I added the association structure, which I forgot to provide for this post.

```
survival_jm_full <-
stan_jm(
formulaLong = list(
Pigmentation ~ Time_num * Food + (1|Fam_Ind),
Length ~ Time_num * Food + (1|Fam_Ind)
),
dataLong = mod_data_longi,
formulaEvent = survival::Surv(Time_num, Survival) ~ Food,
dataEvent = mod_data_event,
assoc = list(c("etavalue", "etavalue_etavalue(2)", "etaslope_data(~ Food)"),
c("etavalue", "etaslope_data(~ Food)")),
time_var = "Time_num",
chains = 4, cores=8, seed = 12345, iter = 1000)
```

- Operating System: Ubuntu 18.04
- rstanarm Version: 2.18.2