Quote:
Originally Posted by
mat82284 
Like I've mentioned before. This poll doesn't work right now. Given time it will work. We only care about when the bulb fails not if its running right now.
Not trying to be a know-it-all, but I want people to know that they shouldn't base their buying decisions on this poll. Even if we got a million votes, the poll still won't work. You have to know the number of still running bulbs because so many people are low-usage users, so it skews the results. The problem with statistics is it's sort of complicated to arrive at a result that even marginally is within a decent error margin. I only know this because I am a programmer and have had to deal with the crappy issue of statistics several times before, it's very hard to do it right. I don't mean to hammer the issue and I'm not trying to make you look wrong, I just don't want to see everyone judging it by this poll.
For instance, if you have 10 people at 1,400 hours that didn't have a bulb pop, they will not vote, yet maybe as many as 50% of the users in this forum use their projector less than 1500 hours a year. Some people buy a new PJ before even replacing the lamp. If you take the number of hours that only people with failed lamps had, the BIAS of average lamp life will always be towards a lower number of hours regardless of how long you run the poll. The only thing this poll can sort of give us, is for people that had failed lamps what the average life was, but that doesn't give us even close to an overall average running figure.
There are other skewing factors and bias in this poll.
My previous judgement on the lamp was only based on conditional and marginal probabilities (if one person had a lamp fail many times in a row), that is a much heavier weighting statistic than a generic poll, but even I admit that is not a definitive factor due to environmental running condtions and bias. For all we know the lamps could last 3000 hours if you run it in optimal conditions.
Let's make an educated guess (and that's all this is), if you add some guessing on what the skewing bias might be, we could say the average lamp life is more likely between 1000-2000 hours, and probably 1500 hours, but that's a very skewed guess, and trying to wait for more results to make the guess more accurate just isn't going to work. It could be 2000 hours easily, even 3000+, maybe higher.
So my suggestion is to create a new poll that asks several different questions, that way you can make it more informative and more accurate. You need to ask several questions in different ways.