Can i still use McNemars test even if a call value is equal to zero? 20 Patients of group A were treated and 0 of them had disease recurrence, whereas between 20 patients of the control group, 2 recurrences were noted. What is the p value? Is there a stat. significance?
I think you don't want to use McNemar's test in this case. Can you explain how you propose to set up this data for this test?
after conducting propensity score matching it is suggested to use the McNemars test instead of Chi square or Fishers exact test. But is small sample size an issue?
Small sample size is definitely an issue. McNemar's test uses the Chi-square approximation with 1 df which requires a minimum of 5 obs. per cell if I remember correctly. see p188 of Higgins cite in the below link for a possible work around..
Best, David
https://www.amazon.com/Introduction-Modern-Nonparametric-Statistics-Higgins/dp/0534387756/ref=sr_1_1?ie=UTF8&qid=1520750599&sr=8-1&keywords=Higgins+J+J+nonparametric
Is the idea that, after propensity score matching, that even though you have 40 real people, you will use only 20 virtual (matched) people in the McNemar test? If so, can you tell us with these 20, if you still have zero counts and in which cells they fall?
If not, can you explain how you will fit this data in to a table that you can conduct McNemar's test on?
https://stats.stackexchange.com/questions/76875/what-is-the-difference-between-mcnemars-test-and-the-chi-squared-test-and-how
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics