This was a system where I was supposed to display a star-rating (1-5 stars, with fractional stars as well) for each item. The people reporting the problems wer

This was a system where I was supposed to display a star-rating (1-5 stars, with fractional stars as well) for each item. The people reporting the problems were the users, and yes, this was bias, as they expected averages. That was exactly my point — while the statistics behind this method are sound, it is not what people expect. Building systems that don't do what people expect is difficult.