Consider a sequence of i.i.d Bernoulli RV's $\{X_i\}_{i \in \mathbb N}$ with parameter $p$. Based on this sequence we build a sequential estimator of $p$ (using inverse binomial sampling for example). This means there's a stopping time $N$ on the sequence, and the estimate $\hat p$ depends on the values of the sequence up to the stopping time, $X_1$, $X_2$, ... $X_N$.

If the estimate depends on some auxiliary random variables independent from $X_i$ (say you flip a coin and based on that you change the third decimal of $\hat p$), I usually say that the estimator $\hat p$ is randomized. (Is that the standard name?)

Now assume that the stopping time $N$ depends on auxiliary random variables too. I'd say the stopping time is randomized. How to call the sequential procedure in that case? If I say "randomized sequential procedure" it may give the impression that $\hat p$, and not $N$, is randomized. I'd like to say something like "randomized-stopping time, randomized-estimator sequential procedure". Is there any term for that, or what would you call it?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.