0
$\begingroup$

I am trying to learn a bit about the analysis of categorical data. I have a problem with one of the assignments.
Drug treatment A was found effective 70% of the time (66%, 74%) in the previous 8-month study. Two new drugs (B and C) have been developed that are claimed will successfully treat the drug within 3 months. A study was conducted to compare these two new drugs. Patients were assigned at random and double blind to take either treatment B or C.
The task is to simply find and elaborate on the best treatment. The results of the study have the following results. 

            y
Treatment   0   1
        B  81  53
        C 100  39   
 structure(c(81L, 100L, 53L, 39L), .Dim = c(2L, 2L), .Dimnames = list(
    Treatment = c("B", "C"), y = c("0", "1")), class = "table")

What is an appropriate approach/test to compare these drugs? Thanks in advance.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
2
$\begingroup$

Assuming that the samples are not biased in favour of any of the drugs, you can try to calculate the 95% confidence interval for each of the new drugs and check whether or not they intersect with the (66%-74%) you had for drug A.

I didn't do the math precisely, but The results will look something like (32%-48%) for drug B and (22%-38%) for drug C. I think a larger sample would really help

Improve this answer
$\endgroup$
3
  • $\begingroup$ I was afraid that it might be that simple. I am just unsure about your estimates because I got them much lower because 0s are non-effective treatments. $\endgroup$
    – m_c
    Commented Feb 24, 2019 at 18:40
  • $\begingroup$ Oops! You are absolutelly right! My confidence interval are of course for the 0 class!! $\endgroup$
    – David
    Commented Feb 25, 2019 at 7:28
  • $\begingroup$ Maybe just edit your answer for future readers. $\endgroup$
    – m_c
    Commented Feb 25, 2019 at 9:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.