[Date Index] [Thread Index] [Author Index]

# tournament statistics

```
For the sake of those who are interested about such things, I am
submitting some raw data that should raise some important issues about
tournament management and presumed affirmative biases (the data will be
reported formally as part of a paper that is currently in preparation.

I have been concerned over the past three years over the role of speaker
points in determining seeding both in prelims and in elims.  I have also
been concerned about the effect of aff/neg skew on side constrained rounds
and in elims.  Consider the following data.

This year, there were 782 prelim debates at CEDA Nats (I'll report on
elims later).  In those debates, the affirmative had a 434-348 record
(55.50% winning percentage - 11.00% difference between aff and neg).  This
compares to a 467-393 (54.30%) record last year and 431-391 (52.43%) the
year before.

Of greater interest is the following breakdown.  For each debate it is
possible to make a prediction as to who wins the debate based on SIDE, W/L
RECORD and/or POINTS.  To accomplish this, one computes an adjusted record
and points that excludes the round for which you wish to make a prediction
(ie. the record and points that the teams received in the other seven
debates).  This creates 18 possible combinations:

AFFIRMATIVE VICTORY

60 	9 SAME RECORD			SAME POINTS

NEGATIVE VICTORY

17	9 SAME RECORD			SAME POINTS

Until the paper comes out, I will leave to your various interpretations as
to what this data means (if anything - tests of signif and multiple
regression will be run later).  I think, however, that two robust
conclusions should emerge.  First, and perhaps somewhat surprising is that
POINTS are a much BETTER predictor of the outcome of the round than
record. Given situations where the team with the higher record does NOT
have the higher points, in 93 of 114 cases, it is the team with the higher
points that prevails.  It is also relatively clear that being aff is a
significant advantage.  In the 77 rounds between teams with the same
number of adj wins and adj points, the AFF has a record of 60-17!

Beyond this, I'll be interested in everyone's observations and the
implications it has for how we run our tournaments.

More later,

Sincerely,

GARY

```

Follow-Ups:

Archive created by Jonathan Stanton (jonathan@cs.jhu.edu)