Originally Posted by jgray91
Now that I read a bit about sampling bias, I still think that as biased as I am with my god tier taste, it still makes sense for 5 = Nami-level futsuu averagely normal average.
Hyperbole and taste aside, someone still should explain sampling bias in a less convoluted way than Wikipedia to me. :P
Most things you measure you do by taking a sample. For example, you can't test something on all the people in the entire universe, so you have to do with a finite number of them, say 1000 people. These 1000 people are your sample.
However it's often impossible to select those 1000 people at random. For example you might get your sample by putting up an ad in the newspaper, but then all those 1000 people will be the readers of that specific newspaper. Usually then, those people have something in common too, which might alter the outcome of your measurement.
Say for example I'm testing how many people smoke out of 1000 people. If those people are all very highly educated I would get a significantly lower percentage than if I'd choose those people at random from the entire population. That is because the sample I used was biased.
In my case I'm basing my rating not on the average quality of all shows that have ever aired, but rather on the average of the shows I have watched. I also base my expectations of how good something should be by comparing it to stuff I have already watched -- not the shows I haven't watched yet. This causes a sample bias, and in my case my expectations of shows keep rising and rising causing me to drop shows at the tiniest hint of not being really good.
As such I would find it unfair to call a 5 average, when really my sample on which I would base that average is extremely biased. Hence I decided to raise the definition of average to 6.5-7 ish.