DZ said:Stephen said:When a study quotes a diffrence of 1-2 percent for groups of 20, I
tend to think that is within a margin of error.These sorts of statements are only true for distributions where the
variance is a well defined function of the mean. Bone density is
probably not one of those.If the true means for 2 normal groups are 99 and 100, with the common
variance of 1, there will be 87% power to detect that 1% difference at
the 5% level, using samples of only 20 people for each group, i.e. 87%
of p-values will be smaller than 0.05.But if both means are precisely 100, then no matter how big the sample
size is, only 5% of p-values will be smaller than 5%.
Which is why I say, "I tend to think" and...
"being a generalization of the thinking itelligent decision making
process, and rational behaviour in general..."
"...so if it's Ok with you I'm not going to shun away from using those
in the future"
The word "percent" is like catnip to you isn't it? LOL!
Stephen - what're the odds?