Stats
Perfect isn't quite good enough (December 12, 2006)
Category: Small sample size issues
Someone wanted me to double check their calculations for Fisher's Exact test. If the
control group, 3 out of 10 patients experienced an unfortunate outcome. In the treatment
group none did (out of 6). You would think that a perfect result in the treatment group would
be compelling, but the one-sided p-value for Fisher's Exact test is 0.21.
That calculation is a bit disappointing, but perhaps not too surprising. The rule of three
states that when you observe zero events in a sample of n patients, then 3/n is an
approximate upper 95% confidence interval. So with 0/6, that upper limit would be 3/6 or 50%
which is well above the control rate of 30%. So perfection just isn't good enough when
you have such small sample sizes.
Related web pages:
07/08/2008.
