Censored likelihoods with different bounds per observation

I’m modeling a problem where the bounds differ by observation (specifically event counts are limited by the a number of opportunities available). Am I reading the docs correctly for the Censored distribution that bounds must be scalar values? I’ve tried passing a vector to the upper parameter but the resulting parameter values are all constant (while the same model, ignoring the censoring bias, produces reasonable results).

Would I need to implement the Censored functionality myself with the CDF of the target distributions?

The bounds can have the same shape as your observed data just fine, they don’t need to be scalar.

Okay, I must have another issue then, thanks for confirming!