[eDebate] Wake Forest judge assignments

Gary Larson Gary.N.Larson
Sat Nov 11 09:37:32 CST 2006


One of the artifacts of the current system is that teams lack the
immediate sense of how well they are doing or not doing in their judge
preferences.  While everyone knew what a 1 or 3 or a 5 was, there isn't
the same sense with the rating data that was provide.

Perhaps the best way in any system is to translate your results into
ordinal terms since even category labels are only a proxy for the
ordinal distribution that underlies them.

For rounds 1 and 2 at Wake we had the following outcomes in ordinal
terms (out of 159 judges)

1-9        56    51
10-19    20    22
20-29    23    19
30-39    14    17
40-49    11    16
50-59    12     7
60-69     2      5
70-79     0      1

To put this is context, had we used 9 categories, there would have been
18 in each category.  So the system essentially had all but 27
assignments in category 3 or better (with the caveat that since teams
have ties in ordinal ranks that might not correspond to category
boundaries, some teams would have more judges in categories than they
would otherwise assign).

But in any case, in aggregate the assignments for rounds 1-2 are as
good or better as could be expected in categorical systems.






More information about the Mailman mailing list