Accuracy vs. precision #529
Description
I'm only working through the first chapter of the book, but some of the exposition is a bit confusing to me. The author writes:
"So two sensors, even if one is less accurate than the other, is better than one."
While this is true, the examples seem to be demonstrating two sensors, one of which is more precise than the other. An accurate sensor would return a reading which is close to the actual weight, whereas a precise sensor would return readings which are spread more tightly around some average.
We've considered the limiting case where the error bars just barely touch, but in the case where the error bars do not overlap at all, what do we conclude then?
I think the 'true weight' of the person would be better expressed as a probability distribution incorporating the information from the two sensors, but the 'reliability', if you will, of the two sensors should also be considered. If I have a scale which consistently reads 10lbs, it may be very precise, but simply incorrect.
Not trying to nitpick, I just don't really understand the gaps I've seen here.
