R help - Nov 2011 - non-parametric sample size calculation

Hi,

I am trying to estimate the sample size needed for the comparison of two groups
on a certain measurement, given some previous data at hand. I find that the data
collected does not follow a normal distribution, so I would like to use a
non-parametric option for sample size calculation.

I found the pwr package but I don't think it has this option and on the
internet found that http://www.epibiostat.ucsf.edu/biostat/sampsize.html says
only PASS allows non-parametric sample size calculations (although the webpage
is not updated).

Any help would be greatly appreciated

Thanks,

Dave
 		 	   		  
	[[alternative HTML version deleted]]

On Nov 3, 2011, at 3:47 AM, David A. wrote:
> 
> Hi,
> 
> I am trying to estimate the sample size needed for the comparison of two
groups on a certain measurement, given some previous data at hand. I find that
the data collected does not follow a normal distribution, so I would like to use
a non-parametric option for sample size calculation.
> 
> I found the pwr package but I don't think it has this option and on the
internet found that http://www.epibiostat.ucsf.edu/biostat/sampsize.html says
only PASS allows non-parametric sample size calculations (although the webpage
is not updated).
> 
> Any help would be greatly appreciated
> 
> Thanks,
> 
> Dave

The first question is how "non normal" are your data? If you used some
formal test for normality and the p value was <=0.05, I would suggest that
you search the R-Help archives for a plethora of discussions on testing for
normality. You will find that such tests should largely not be used in deference
to the question "Are the data normal enough?". If they are or can be
transformed reasonably, use standard functions for calculating power and sample
size, such as power.t.test().

If you need to use a non-parametric test, you might want to review this page by
Jerry Dallal:

  http://www.jerrydallal.com/LHSP/npar.htm

which has some general guidelines for calculating sample size predicated upon
using standard parametric tests and then adjusting the sample size using the ARE
(asymptotic relative efficiency) based upon the non-parametric intended.

HTH,

Marc Schwartz