On Thu, 08 Jun 2000 12:53:34 GMT, Mithat Gonen <[EMAIL PROTECTED]>
wrote:
> In article <5PX_4.402$[EMAIL PROTECTED]>,
> "Isabelle Gaboury" <[EMAIL PROTECTED]> wrote:
...
> > I'm looking for articles and books (or authors) which can help me to
> > compute the sample size (and related power) in case of a case-control
> study
> > of a rare disease.
< ... >
> Assuming you outcome is also categorical (like disease status yes/no,
> or present/absent), I would start with Fleiss's book "Statistical
> Methods for Rates and Proportions" where he has introductory sample size
> discussions for a variety of situations.
>
> In a rare disease situation, the number of cases will be limited and
> hence it might be more practical to calculate the minimum detectable
> difference, i.e. start with a feasible sample size and a reasonable
> power (80-90%) and calculate the minimum difference you can detect with
> this sample.
>
> Mithat
- I know how to state a "minimum difference you can detect" with a
sample, but I don't know what Mithat has in mind for the 80-90% power
part of the question.
In general, you can answer a lot of complicated power questions, to a
good approximation, by providing the near-equivalent 2x2 table of
outcomes, or a t-test between two groups.
When you dummy up your table, you can compute your statistical test.
A chi squared test statistic will increase in proportion to N; most
tests increase with N or sqrt(N), so you can project from a single
dummy-example to see what the "detection" point is, where the N gives
you the test-size you are looking for -- 3.84 is exactly the
chi-squared, for instance, for the 1-DF test at 5% -- your "power" is
50% at that value of N. That is, on replication, you expect to do
better or worse that this, even odds.
For a 1-df chi squared, doubling that N is what is needed to increase
the power to 85%.
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
