0

I have been using the Wilson score interval to determine whether the registration rate from two different website flows are significantly different. I want to see if there are suggestions for better approaches for small samples of binomial distributions.

BradHanks
  • 111
  • Is your primary interest in comparing two different website flows, or in knowing the confidence intervals per se? Also, are your data simply the number of total visitors and the number of registrations for the two flows? – EdM Jan 20 '15 at 20:00
  • 1
    Isn't the Wilson score interval for a single proportion? – Glen_b Jan 21 '15 at 02:48
  • 1
    I am interested in knowing if the two conversion rates are statistically different. So I was using the Wilson score to calculate the two intervals and then see if they overlap at all. – BradHanks Jan 21 '15 at 17:45
  • What do you mean by small samples? In any case, have you considered a z test of two proportions? – Joel W. Jan 21 '15 at 22:37
  • So it wouldn't meet this condition -> Each population is at least 10 times as big as its sample. I guess I could just take a sample of the population, but with colleagues that have an even lower comfort level with stats I am trying to find something that would work with two populations. – BradHanks Jan 22 '15 at 18:12
  • For the relationship between tests and confidence intervals, see https://stats.stackexchange.com/questions/18215/relation-between-confidence-interval-and-testing-statistical-hypothesis-for-t-te – kjetil b halvorsen Mar 09 '23 at 18:38

0 Answers0