topic Sample size calculation: noninferiority design with paired proportions in Statistical Procedures

Sample size calculation: noninferiority design with paired proportions

AnalytX — Thu, 26 Jul 2012 07:54:03 GMT

Dears,

I would like to compute sample size calculation for the following study design:

. Nonferiority test with the following tolerance margin: Proportion 1=0.15 and the acceptance margin allows to obtain Propotion 2=0.20

. Statistical power: 0.8

. Type I error: 0.05

If I am not mistaken, thare are very few examples about the PROC POWER with PAIREDFREQ statement available.

Do you have an idea how to calculate the sample size using these hypotheses?

Thanking you in advance for your help. It will be very helpful.

Regards,

Re: Sample size calculation: noninferiority design with paired proportions

AnalytX — Fri, 27 Jul 2012 07:52:39 GMT

Dears,

Please find below what I have tried but I am not sure about the different options to use (what is the better hypothesis fo corr, what is the most adapted approximation method, ...) :

proc power;

pairedfreq test=mcnemar method=connor dist=normal

alpha=0.05

sides=U

pairedproportions = (0.2 0.3)

corr=0.5

power=0.8

npairs= .;

run;

Thanks in advance for your suggestions or comments.

Regards,

Re: Sample size calculation: noninferiority design with paired proportions

AnalytX — Thu, 02 Aug 2012 07:27:11 GMT

Dears,

No idea about this topic?

Regards,