[peregrine]

162 hardly seems like a good enough sample to make any certain conclusions...not saying his data are any better just pointing it out.

[rjprins]

This is a common misconception. You want to calculate the odds of 162 people not showing an effect where there should be one?

What is important is not the sample size, but the manner of sample selection. 10 people carefully chosen to represent a population is infinitely more valuable than 1000 people chosen with some uncontrolled selection bias.