# Re: st: RE: RE: Samspi non normality?

 From Jeph Herrin <[email protected]> To [email protected] Subject Re: st: RE: RE: Samspi non normality? Date Fri, 10 Nov 2006 13:49:21 -0500

```Paola,

Stefano has pointed you towards a very good resource. In
distributions and want to compare means, one simple if
conservative way to proceed is to test using a Wilcoxon
rank sum test. If you use this test, then you can get
a good estimate of the sample size by using -sampsi-
and then dividing N by the Pitman A.R.E., 0.864. That
is, if -sampsi- says you need n=600 for a normal distribution
then you need (no more than) n'=600/0.864 ~ 695 for a
rank sum test.

Hope this helps,
Jeph

Di Stefano, Paola wrote:
```
Dear Jeph and Roger,

Thank you for your precious help.

My hypothesis is:
H0: mean (treated group) < mean (control group) But I don't know anything about distribustion.

Ciao
Paola

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Newson, Roger B
Sent: venerd� 10 novembre 2006 18.54
To: [email protected]
Subject: st: RE: Samspi non normality?

As Jeph says, it depends on what kind of non-normality you are going to assume. A common alternative is the lognormal assumption, although I don't know if that is what you want to use in your case. If we think the data are lognormal within groups, then we test for detectable geometric mean ratios, not detectable arithmetic mean differences.

You might like to read my paper on the -powercal- package (Newson, 2004), which discusses power calculations for a wide range of non-normal cases, including lognormal geometric mean ratios, and even detectable levels of Somers' D when rank methods are used. The -powercal- package can be downloaded from SSC using the -ssc- command in Stata. If you do not have access to The Stata Journal, then you can download a pre-publication draft from my website (see my signature below).

I hope this helps.

Best wishes

Roger

References

Newson R. Generalized power calculations for generalized linear models and more. The Stata Journal 2004; 4(4): 379-401. Download pre-publication draft from my website at www.imperial.ac.uk/nhli/r.newson/

Roger Newson
Lecturer in Medical Statistics
Respiratory Epidemiology and Public Health Group National Heart and Lung Institute Imperial College London Royal Brompton campus Room 33, Emmanuel Kaye Building 1B Manresa Road London SW3 6LR UNITED KINGDOM
Tel: +44 (0)20 7352 8121 ext 3381
Fax: +44 (0)20 7351 8322
Email: [email protected]
www.imperial.ac.uk/nhli/r.newson/

Opinions expressed are those of the author, not of the institution.

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Di Stefano, Paola
Sent: 10 November 2006 16:47
To: [email protected]
Subject: st: Samspi non normality?

Dear all,

I'm trying to estimate a sample size for two-sample comparison of means using the command sampsi.
The study is to test for a difference in effectiveness of 0.5 between two groups, assuming a sd(1.4) Is there any way to change the normality assumption? Is there another way to estimate sample size?
Thanks
Paola

*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/

*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/

*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/

```*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```