I always thought we should have several averages: Full data, cut off 1% at each end, 5% at each end, and 10% at each end. I think we'd get much more out of th

I always thought we should have several averages: Full data, cut off 1% at each end, 5% at each end, and 10% at each end. I think we'd get much more out of the A-1 or A-5 than the raw average. (Of course, the real answer -- like with the weather -- is that you need to use distribution information, not a single value.)