Okay this is getting ridiculous. Some people here clearly know nothing about polling and nothing about statistics. I'm not even going to get into speculating the failure rate, im just going to explain to you why poll's on websites are largely useless for anything:
I'm going to start this of by a little example:
Have you ever seen a political debate show where people can call in and vote on stuff? Like say "Legalize abortion, yes or no". Now, some of these programs are very popular, getting hundreds of thousands of callers during a show. In this particular example, a program in Norway named Tabloid, asked the question if the government should stop allowing so many immigrants to settle in Norway. Shockingly 80% of the callers (50,000 total callers (total population in norway is 4.5 million) said that we had enough immigrants as it is.
Now on the day after this show was on, a analytical firm that does polls, had also done a poll on the same issue. They got a 30% number for the "no immigrants people", they only polled 3,000 people though. Guess which one is more scientifically accurate? You guessed correctly, the smaller, professional poll, because the people who watch "Tabloid" on a friday night and call in, aren't representative of the general public.
In order to have statistical data of ANY scientific value you need to make sure that the people who are responding in your poll are representative as an average of all the consumers who are using this product. Not only that, but these polls also need to have the "right" questions, you can easily manipulate the results of a poll, by just slightly changed the question or the answers.
We on the same page so far? Good.
Now, lets take at the polls your discussing. Are the people answering this poll representative of a the average gamers? No they are not. The most people who own this console are not spending their time on forums. Further, the average gamer will probably be using his console very differently than the average guy at a gaming forum. They will most likely spend less time playing, this alone leads to a lower score for the reliability than it should be. (Because obviously, since the average player plays less than the forum voter, they will most likely have consoles that are still alive).
Not only that, but the results can also largely be flawed, because of people lying, as we know in this generation fanboy wars on forums are taken to the next level. It happens everywhere, even here on beyond3d (tho to a lesser extent, but its still easy to spot biased posters [they usually have a little red dot below their name]), anyways, nobody is preventing these people to vote as well. And if you look at the amount of people who talk about how much better GT5 will be over Forza 2 at the Forza 2 forums on GameFAQs, there is little doubt that some people will post "negative" values to consoles they have never owned.
Further, are we sure the pollers are voting correctly? Most of these questionares are "Has your X360 broken down?" Now, just take a look at the B3D poll, you can already tell its flat out wrong. Why? Because most people are only voting once, they vote "Yes my x360 has died". They dont vote that their current X360 is alive and well. And 99,9% of everybody who has had a dead x360 has gotten a replacement (granted shifty did want people to vote for both), but if you cannot vote for both, how on earth are you going to get anything that is remotely accurate?
Surely, if you had a x360, and it dies, and you receive a replacement that works. Thats 50% reliability rate, NOT 100%.
Because of all these facts, using website polls to make any real life conclusions is well... to put it bluntly, its stupid.
I learned my one new fact for the day so I guess I can go back to sleep now.
