True, about specific combos. But, just assume it was the same
combo, e.g., 2-3 - the logic is the same. In fact, it was my 2-3
analysis that led to this question.
Median can be a valid method, though a bit more difficult, computationally,
since the series must be ordered.
I'm not too quick to throw out outliers, 'cause, hey, they did
happen. If you take standard deviation into account, outliers
will automatically widen the confidence limits for your 'ultimate'
calculation (whatever that may be).
