Two worthy points there.Zardoz wrote:[...] Polls make matters even worse, because they have the veneer of legitimacy but in fact have none, other than accurately representing the opinions of the people who respond.
That being said, whether a view represents a majority may be less important than whether it represents a significant number, majority or not.[...]
I have no great love of forum polls, other than for very simple queries or for amusement. Most serious issues have too many shades of grey to be answered through a pick-list and the real answers usually lie in the words of the reply posts and not the poll itself. And as you note, the poll is only of those who care to respond. Which leads nicely into the second point.
In any largish population, getting an outright majority view of the whole is likely a forlorn hope, hence you do need to consider when something is expressed by a significant number of people. But that, to me, opens up the question of what is "significant". On a "gut feeling" basis, 10 people out of 50 is going to be significant, but is 10 out of 500? Or 10 out of 5000? I'm sure there's a recognised way of establishing "statistical significance", but I don't know what it is; this probably ties in to Semplerfi's topic on QA and QC - maintaining consistency.