|
|  |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Re: power analysis for panel data
On Nov 7, 2006, at 12:57 PM, Christopher W. Ryan wrote:
The lack of any statistically significant beneficial effect of any
of the interventions on BMI does not surprise me, given the
generally intractable nature of obesity. The general futility of
simple office-based exhortations to lose weight is part of my point.
But what is the power of this study? I don't know how to calculate
that. Am I failing to see statistically significant beneficial
effects on BMI because of inadequate power?
You're right to be concerned about power here (of course there are
other issues too such as the apparent non-random assignment to
treatment, but I'll assume you have already thought carefully about
those). However, power per se is a frequentist concept that really
only applies prior to collecting the data (or at least to analyzing
them). Given that you have already done the analysis, and assuming
you do not plan to collect any more data, I'd suggest recasting the
issue in terms of precision. In other words, ask yourself: Do the
confidence intervals include what would be considered substantively
meaningful effects? If so, then one could argue that the study was
underpowered. If however the confidence intervals are narrow enough
to rule out substantively meaningful effects, then you can claim that
your data actually provide evidence against the existence of such
effects.
-- Phil
*
* 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/
