I am conducting a literature review and would like to compare a single proportion to a group of other proportions.
For example: 7% of patients died using a new method; 21%, 17%, 13%, 6%, 12%, and 7% died using an old method (each proportion in from a different population). I want to do a statistical test to see if the proportion using the new method is different than the proportions using the old method.
Would the Crawford-Howell test be appropriate to use? I think it's usually used to compare a test patient to a group of control patients so I'm not entirely sure. Is there a different test that could be used for this comparison?
Thank you so much for your advice!
No, Crawford-Howell is equivalent to t-test, which cannot compare percentages since they are not normally distributed. Instead, you use a Chi-squared test :)
Andrew Paul McKenzie Pegman
Thank you so much for your answer. I know I can use a chi-square test to compare a single proportion to another proportion. example: 7% vs 21%. But can I compare 7% to ALL of the other percentages in one test. Example: 7% vs (21%, 17%, 13%, 6%, 12%, and 7%)
Yes, you can do multiple comparisons in Chi-squared tests and then test for cell significance :)
- Badges
- Science topic
- Similar topics
- Cognitive Science
- Thinking