One of the conditions for a Bernoulli trial (and by extension binomial proportion confidence intervals) is that the probability of success is the same every time the experiment is conducted.

In the case of a process that has some degree of unavoidable variability in it (e.g. p = 0.8 $\pm$ 0.1), is this still a Bernoulli trial (and can binomial proportion confidence intervals still be calculated)? I understand that if the outcome is yes / no, success / failure, then by definition it is a Bernoulli process, but I am unsure of how to deal with the changing probability of success. The mean stays the same between experiments, but there is variability around it - is that okay?


I'm looking at material breakage. My samples are similar sizes, but have slightly different shapes (like gravel).


1 Answer 1


You might be interested in the beta-bernoulli distribution. This is a compound distribution, where the probability of success is modeled as some beta distribution.

$$ p \sim \text{Beta}(b,a)\\ x \sim \text{Bernoulli}(p) $$

This seems appealing for your problem because the probability of success is explicitly modeled as random.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.