Quote:
Originally Posted by Donger
Well, the 50% compliance was clearly a guess, as it had to be, but apparently the 90% is known with much higher certainty, no? I'd like to find out how they got that figure, though. I thought I read something about mobile phone tracking, but that's a can of worms that I'd rather not open at this stage.
|
My understanding is that it is not based on any hard data at all. It is based on going back over the data and finding a figure that fits the reality. Actually it's a whole set of figures that are adjusted to make the models' algorithms fit the data.
You start with an educated guess and adjust as more hard data comes in. The problem is that (as shown by an alternate model in the UK) you can have vastly different models that fit with the limited data we have at this point.