Close, but you need to do a two-sample test. Wolfr
Post# of 155128
(Total Views: 1549)
Posted On: 06/07/2020 6:00:07 PM
Close, but you need to do a two-sample test. Wolfram Alpha has one you can play around with: wolframalpha.com/input/?i=two+binomial+distribution+test
Enter 0 for the "hypothesized parameter" to indicate the null hypothesis that there is no effect of leronlimab. Then enter the mortality data for the control group as sample 1 and leronlimab data for sample 2. There's a link below the data form for toggling between entering the number of fatalities or the proportion and a "Compute" button below that.
A caveat: this test is based on a normal approximation to the binomial, which works well when the number of fatalities in each group is not small (n >= 5 as a rule of thumb) but kind of falls apart when the number of fatalities is small. It most likely is not exactly the same test that they'll use for the analysis, but it will be in the same ballpark for sure (except possibly for the small numbers, where it might be quite a bit different).
Enter 0 for the "hypothesized parameter" to indicate the null hypothesis that there is no effect of leronlimab. Then enter the mortality data for the control group as sample 1 and leronlimab data for sample 2. There's a link below the data form for toggling between entering the number of fatalities or the proportion and a "Compute" button below that.
A caveat: this test is based on a normal approximation to the binomial, which works well when the number of fatalities in each group is not small (n >= 5 as a rule of thumb) but kind of falls apart when the number of fatalities is small. It most likely is not exactly the same test that they'll use for the analysis, but it will be in the same ballpark for sure (except possibly for the small numbers, where it might be quite a bit different).
(0)
(0)