I have been using channels to compare the effect of different ad positions on CTR etc.
I have been running the comparison on a random sample of about 200 pages for each channel, for about 3 weeks. Page impressions on each channel are about 600, and CTRs vary by a factor of 4.
So it appears one ad position is much better than the other ones.
A long time ago in a distant galaxy I used to be OK with statistical tools, but now I can't remember what technique I could use to see if my results are statistically significant. I.e. I have run the tests long enough to prove say "80% of the time channel A has a higher CTR than Channel B"
Please can someone remind me what statistical technique I could use to do this?