No, it's not that hard to believe - but what i'm saying is that a 20% sample size is statistically good enough to indicate overall QC issues etc. If the people on these boards and Facebook have widespread QC issues with some figure, then statistically it is almost certain that the remaining 80% will also have a similar 'lemon' rate.
It is simple stats really
Granted it's been a while since
I've been in a statistics class, but I remember enough to know what you're saying is not correct.
First, the vast majority of people go online to spread bad news. It's simple human nature to spread bad news. It's taught in every marketing class. This means the vast majority of people online discussing issues will be spreading bad news. The people online are not a valid sample of the entire population because of this. Every statistician will tell you that this is sampling bias.
If the people that go online represent 20% of the market segment and 80% of those people express disappointment with the product, the best you could estimate is that 16% of the market segment is disappointed, not that 80% of the entire segment is disappointed. The online community is not a valid representation of the market. There are still LOTS of 1/6 collectors that don't go online.
You look at something simple, there were 750 of that Rogue PF made, how many people _____ed in that thread?? Maybe 50 people, which is a lot, but it's only 6% of the total edition size. Only 6% of the buyers of that specific statue voiced their opinion online, 6% is not a valid sample size
Look at this
https://www.surveysystem.com/sscalc.htm
Even at lowest acceptable confidence level, 95%, you'd need 254 people reporting to determine a valid sample at 750 units. And there most definitely were not 254 unique voices in that thread complaining.
The people in this forum have IMPOSSIBLY high standards. ABNORMALLY high standards, even.