1

I have fundraising data on two different types of groups, let's call them Group A and Group B. I am looking to perform a hypothesis test on whether organizations in Group A raises more money from Source X as a percent of their total fundraising compared to organizations in Group B.

My initial idea after dong some reading was to perform a two-tailed t-test comparing the proportions of organizations in Group A to that of Group B. However, I have seen some other posts saying that a binomial test would be more appropriate since the data is bounded by 0 and 1. However, I am not measuring "successes" in any real sense.

Is it still appropriate to test this hypothesis with a t-test?

Thank you all for your help!

ModalBro
  • 143
  • 4
  • perhaps this answer might be useful: https://stats.stackexchange.com/questions/136311/t-test-on-percentage-values-among-experimental-replicate-values – william3031 Aug 13 '19 at 22:40
  • Why would comparing the proportion of organizations in the two groups tell you anything about the amount of money that resulted? You seem to be testing something different from what you said you were trying to find out – Glen_b Aug 14 '19 at 03:40
  • @Glen_b: Sorry, I may have not been clear. I am looking at the proportion of money raised from Source X over total money raised for organizations belonging to Group A and Group B. – ModalBro Aug 14 '19 at 16:53
  • How are you computing the denominator in either test? – Glen_b Aug 15 '19 at 01:30
  • @Glen_b: Total amount raised.This is fundraising data from the FEC. – ModalBro Aug 16 '19 at 00:43
  • Not the denominator of the proportion, the denominator of the test statistic. Please give full details of what you actually calculated. I'm concerned that you may be treating continuous proportions as if they were count proportions. – Glen_b Aug 16 '19 at 07:39

0 Answers0