Patryn of Elvenshae said:
Yes, there indeed is the statement that refutes your claim. Please remember, you didn't qualify your assertation along the lines of "most commonly used" (although I have myself addressed this.) What you said was
every other poll in the history of statistics has every been done. Now, this clearly isn't true. If you want to say that complete polling isn't common, you'd have a different point and I'd agree with you. That is not, however, at all what you said.
The only way a full-demographic poll works is if you're targeting a group of less than a hundred members or so.*
Again, not true. As perhaps the most common example that comes to mind, we do a lot of medical research. We have found ourselves polling all of a given medical specialty in Canada. On occassion we get them 100% completed, although all come close because the specialists have a vested interest in answering. These polls are done with over 100 respondents. This is done by arranging in-office interviews, scheduling times to call rather than blanket calling, mailouts with a long window of returns, etc. EDIT: All these methods still use the same standardized, quantitative poll but recognize that just calling people up and hoping for the best won't get it done. Again, it's not common but it certainly does occur despite your claims to the contrary. It's not like we tell a client "what, you want to ask 101 people this survey? Well, I'm sorry, but the best we'll be able to do is get 100 replies for you." The greater the sample, the less the chance of a 100% return, but it
ISpossible.
Seriously, though, are you just being obtuse, or do you not recognize hyperbole? Calling their methods into question because they used the most widely-accepted method of polling is coming it a little high.
Obtuse? Actually, I'm being very direct.
You'll also note I'm not saying that it was their polling that comes into question (I don't have any information on their call plan to make that comment), so much as the conclusions they drew from it. There are some conclusions in there (such as the # of gamers in the US) that simply cannot be arrived at with accuracy from such a poll--certainly not one instance of such a poll. To even arrive at a reasonable estimate they would have to conduct a longitudinal poll of shifting demographics to see if patterns of consistency showed up throughout the various regions. Without doing so extrapolating a nationwide number is an guess the data simply can't support truthfully.
Joe Average? I'm not here to compare resumes, bub.
That much, sir, is obvious.
It would be nigh-impossible in any "real world" application, however.
Say with, oh, I don't know, medical specialists?