two interesting statistics

[Archive]

Posted By: davidcycleback

Two other things I noticed.

From my feedback at least, I see that only a small minority of buyers are leaving feedback, much smaller than before. This presumably is because buyers get a 100% rating no matter what, so there's no incentive to give and receive feedback. This means that the feedback statistics are not representative of what the buyers but what a small portion left. It also means one buyer with a vendetta could lower the seller's rating significantly by giving the lowest ratings across the board. 5 + 5 + 5 + 5 + 1 = 4.2 It is noteworthy that proper statistical 'scoring' methods automatically drop the high and low scores to remove the inevitable aberrations or freak numbers-- just as they do in Olympic skating before calculating the final score.

The other nonsensical thing is that an eBay feedback score below a 4.2 will result is suspension, while the buyers giving the ratings are told that a 4 means you are satisfied with the seller's performance, that the seller's performance was acceptable. That's the equivalent of the Dean telling professors that a B is a passing grade, while telling the students that if they receive a B they will flunk out. A student can by flunked out by a passing grade.

There's do doubt it's hard to make a good in practice feedback system, but a start is to removing obvious Catch-22s and Alice and Wonderlandisms. "4 is a passing grade, but if your grade is 4 you fail" is Catch-22 exemplified.