|
Kantor said: But we don't know that this sample is accurate like you described. For all we know, half of the Wii's could have been used less than an hour a day, and only 10% of PS3s and 5% of 360s. And there are several factors in the equation- usage time, storage conditions, country, care when handling, etc. We don't know that this sample is any good. I mean, the title could at least state that this was a survey of 16,000 people. I also highly doubt more 360s succumb to disc read errors than RROD... |
On paper, yes those are concerns. In reality it doesn't really work out that well. The probability of getting a terrible data set like you described is virtually non-existant if the data is an accurate sample. I wish I could show you the proof of it, but thats how it comes out in the end. If you want to try and run the numbers yourself you can try to think of it as a bag of marbles. Take 100 million marbles, and divide them up into whatever broad categories you want. Then calculate how likely you are to get enough of a certain type of marble to severly influence the average population. It gets to be an absurd improbability at an extremely rapid pace.
A short hand example that won't help much. Lets say you have 3 green marble, and 97 blue ones. If you picked 10 marbles at random, how likely are you to get a large number of green marbles? You could get all 3 in theory. Assuming no replacement the propbablity would be 1/(100*99*98) or 1/970,200. To get just 2 marbles in would be 1/9900 which is significantly more likely, but still extremely improbable. Hopefully this gives you some idea of how it scales. You can be off by a little bit, but to be way off becomes almost impossible.
