^^^ Averages can be locally accurate, and still reflect an overall accuracy as well...people do averages all the time on a corporate scale that includes only a very small ACTUAL number of people in the overall user base. If you look at the claims most carriers use as far as "the most preferred" or "the best working", these are usually based on a very small fraction of user base that was asked and gave answer...this is no different, it was just a way to come up with an average, obviously the entire iPhone community would not be here to participate so the average would not reflect such, but it does reflect a community, and that in itself makes it legitimate.
Studies have found that 100 people asked at random can, in most cases, reflect a very accurate picture of a much larger average pool...the numbers will obviously skew one way or another but it is the pattern that begins to form in such a small and local poll.
Take Nielsen for instance...obviously a VASTLY larger poll group, but the numbers seem to follow our own fairly accurately with the exception of the older group, who obviously are not going to be as involved in online forum posting and communities, thus making their numbers drop in comparison with neilsen's.
This nielsen poll was also taken before the $199 iPhone 4 was released, which in itself would yield higher numbers in the younger age base...increasing both the teen/early 20 and mid 20/early 30 focus group, which deff reflects what our poll shows with just the forum responding.