@thread
It's nice to read all the opinions and experiences in this thread, quite a variety 
I will apply some statistical analysis to the results once the poll runs it's course at the end of the month. A stats professor at my University has written an excel spreadsheet for me that will do this.
He said to me if you estimate a total population size of 2000, 5000, or 10000 possible users, the poll has a plus or minus accuracy rate of 9.5%.
Extrapolating those numbers out based on the population estimate could give you an estimate of possible sales.
For example, currently the poll has 38% of respondents saying "Yes!".
38% of various estimated population sizes is:
For 2000 Users: 760
For 5000 Users: 1900
For 10000 Users: 3800
The variance is plus or minus 9.5% as I stated.
At least that is how I interpret what the Professor has said to me. But once the poll is finished I will ask him exactly how to interpret the results and I can even post the spreadsheet here. 
I'm by no means a stats expert so I could be totally wrong.
I don't think that is correct. The so-called standard error is the square root of [p*(1-p)/n],
where p is the probability (0.38) and n is the sample size. Using the standard error gives only a 68% confidence of the result. For a 95% confidence, you use ~twice the standard error. I have no idea how many have participated in the poll, but let's say it is 100 (for example). Then the standard error is about 5%. So at the 68% confidence level, that would be 38% +/- 5% and at the 95% confidence level, that would be 38% +/- 10%. BUT THE REAL PROBLEM IS you're assuming that the AW website is a good representation of Amiga users. If it is tilted toward OS4 fans, then extrapolating these results to all Amiga users is not a good idea.
