intrepid2566, page-24

Kogarahboy,

The problem is that you raise a some potential scenarios, including one you mention where....

"I would be hoping to see that in at least a few cases we had a strong positive response to the drug."

Let's accept that that does indeed happen (not certain of course). Statistically, that would be invisible within the

'100 active doses' you mention later. Of course, the real effect might be greater, or less and the extent of each effect might be substantial or relatively minor. In other words, the possibilities are so many that it is impossible to look at unblinded data and reach a conclusion. This is not a yes/no situation where you are starting at zero. Some patients will show some improvement without drugs, some wont and the extent of improvement will likely vary a lot between patients and the actual results will be influenced on how the parameter that is measured. All those variables means that you (and the docs) can't just look at the data from unblinded groups and figure out what is happening. You can't just pluck out the 5 'top' scores from a blinded group of 100 and assume they must be part of the treated group. notr can you look at some stable patients and asume they are in the control group. Likely they are a mix of both and you'll need unblinding and heavy duty stats to determine if there is a real effect.