Stan samples/simulates from parameter the same way in 2 models?

Hi I’m working with a physiologically based kinetic model where I’m comparing one compound to the other by dividing their area under the concentration-time curve (auc)
I code my codes using stan, but realize that for the output to be logical, I need to ensure that I’m dividing auc from individuals with the same physiology. For example, in one compound model, the auc I get for later division by the other auc from the second compound model should be from individual with low excretion (not low in one model, but high in the other model, as currently sampled with stan because it is random in a sense). Is there a way to fix the way stan sample from one parameter in two parallel models? Or random sampling is just the current state of the art?

Many thanks,
Frances

Howdy!
It would help if you could include your Stan code, but just based on your brief description it sounds like you might could write the generated quantities section such that you only carry out the computations for certain individuals.

I’m not sure what you mean by this. You can have the same parameters in two (or more) likelihood statements, for example, like:

   for (n in 1:N) {
      target += bernoulli_logit_lpmf(Y[n] | a + r_1_1[J_1[n]] + r_2_1[J_2[n]]);  // model 1
    }
    for (n_2 in 1:N_2) {
      target += bernoulli_logit_lpmf(y_c[n_2] | a_b + (inv_logit(a + r_1_1[K_1[n_2]] + r_2_1[K_2[n_2]]))*b);  // model 2
    }

where the inverse logit of the linear predictors in model 1 become a ‘predictor’ in model 2.