1
$\begingroup$

I am going to run a test with N subjects. There are M=50 stimuli to be evaluated by the subjects but, for time and concentration reasons, a subject cannot evaluate all the stimuli. Therefore, I am going to provide only P < M stimuli (randomly chosen) to each subject.

  • How can I compute the expected number of times a given stimulus will be evaluated, knowing N and P?
  • Similarly, how can I choose N in order to have an expected number of evaluation greater than X for each stimulus?
$\endgroup$

2 Answers 2

1
$\begingroup$

If all the stimuli are assigned randomly (uniform) and independently, then subject $n$ would have probability $\frac{P}{M}$ to get a specific stimulus $p_i$ in his set of stimuli $P_n$:

$\mathbb{P}(p_i \in P_n) = \frac{P}{M}$

To get the expected number of times that stimulus is given in total, simply multiply by $N$:

$\mathbb{E}(\sum_n 1_{\{p_i \in P_1, \dots, P_N\}} = N\frac{P}{M}$

Similarly, if you want $N\frac{P}{M} > x$, you can rearrange this to find:

$N > x\frac{M}{P}$

$\endgroup$
1
$\begingroup$

This is just a binomial distribution.

A given stimulus has a probability P/M to be evaluated by a given subjects. Therefore, the number of times an stimulus is evaluated in a N subjects sample follows a binomial distribution $B(N,P/M)$ and its expectation is $N·P/M$. To have an expected number greater than X, you just need to choose $N>X·M/P$.

$\endgroup$

Your Answer

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

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