Thank you David and Ronan. I'm confused by the
apparent contradictions in the two posts. I understood
how Ronan calculated the power but its this valid? Or
is the problem (as I originally suspected) that a
"difference of 1.5 SD" doesn't mean anything without
knowing something about the distribution of the two
groups? Am I missing something?
Thanks in advance,
Ricardo.
--- David Moore <[email protected]> wrote:
> Ricardo,
>
> You're correct, but the "other" statistician
> probably made a number of
> assumptions to fill in the blanks, so your colleague
> wouldn't be bothered
> with having to think about the substantive issues.
> I did statistical
> consulting for several years and this (power
> analysis) was one of many
> problems that I had to spend a lot of time
> explaining to clients. Most of
> the time I ended up walking them through sections of
> Cohen's seminal book
> just to save myself from going insane. I suspect
> the pathologist's friend
> was using Cohen's (cook)book or a software package
> based on Cohen's formulas
> and suggestions. If he was a true friend, he would
> have spent the time
> explaining the underlying logic to the pathologist
> and he (the pathologist)
> would have been able to speak intelligently about
> his expectations for the
> means and standard deviations.
>
> It's been my experience that many experienced and
> fairly competent
> researchers do not really conduct a proper power
> analysis, when it's called
> for. Your experience with the pathologist is a
> prime example of the
> investigator proceeding without having a full
> understanding of the problem
> at hand. In my opinion, if he doesn't have a clue
> about the effect size of
> interest, he shouldn't be doing the study.
>
> Dave
--- Ronan Conroy <[email protected]> wrote:
> on 18/02/2003 5:11 pm, Ricardo Ovaldia at
> [email protected] wrote:
>
> > and furthermore I did
> > not need this information because his friend
> (another
> > "statistician") told him that with 8 observations
> per
> > group he could detect a difference of 1.5 standard
> > deviations with 80% power.
>
> . sampsi 0 1.5, sd(1) n(8)
>
> Estimated power for two-sample comparison of means
>
> Test Ho: m1 = m2, where m1 is the mean in population
> 1
> and m2 is the mean in population
> 2
> Assumptions:
>
> alpha = 0.0500 (two-sided)
> m1 = 0
> m2 = 1.5
> sd1 = 1
> sd2 = 1
> sample size n1 = 8
> n2 = 8
> n2/n1 = 1.00
>
> Estimated power:
>
> power = 0.8508
>
> A different of 1.5 standard deviations is very large
> indeed. Yes, the power
> is 85%, but the Mann Whitney statistic is 86%
>
>
> . mwstati 1.5 1 1, i
> Mann-Whitney statistic for this situation is: 0.8556
>
> That is, the probability that an observation in
> group 2 will be higher than
> an observation in group 1 is 0.8556.
>
> Two comments:
> 1. The minimum clinically significant difference in
> Cox2 may well be less
> than 1.5 SD and
> 2. An 80% power is an unacceptably high risk of
> failure. I don't get out of
> bed for less than 90%.
>
>
> Ronan M Conroy ([email protected])
> Lecturer in Biostatistics
> Royal College of Surgeons
> Dublin 2, Ireland
> +353 1 402 2431 (fax 2764)
>
> --------------------
>
> *
> * 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/
=====
Ricardo Ovaldia, MS
Statistician
Oklahoma City, OK
__________________________________________________
Do you Yahoo!?
Yahoo! Shopping - Send Flowers for Valentine's Day
http://shopping.yahoo.com
*
* 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/
