Here's the numbers again where spread >= (1/4)(total). I added two extra columns. The extra columns attempt to capture the correlation between the events. "Fav" shows the win/loss/tie record of the over given that the favorite covers. "Dog" shows the win/loss/tie record of the under given that the underdog covers.
<font class="small">Code:</font><hr /><pre>Season Over/Fav Under/Dog Fav Dog
2007 6-7-0 2-10-1 6-2-0 2-2-1
2006 10-27-1 9-28-1 10-6-1 9-11-1
2005 5-19-0 8-16-0 5-5-0 8-6-0
2004 3-13-1 4-13-0 3-2-1 4-7-0
2003 4-15-0 6-13-0 4-5-0 6-4-0
2002 6-16-0 4-18-0 6-5-0 4-7-0
2001 5-16-0 2-17-2 5-6-0 2-6-1
2000 8-20-1 6-22-1 8-8-0 6-5-0
1999 6-20-0 8-17-1 6-5-0 8-6-0
1998 8-18-0 4-22-0 8-5-0 4-9-0
1997 3-17-1 4-16-1 3-5-0 4-7-1
1996 6-16-1 4-19-0 6-6-0 4-6-0
1995 4-21-0 9-16-0 4-3-0 9-9-0
1994 9-15-0 9-15-0 9-3-0 9-3-0
1993 12-31-0 15-28-0 12-7-0 15-9-0
1992 8-44-1 15-38-0 8-9-1 15-20-0
1991 8-23-0 7-24-0 8-8-0 7-8-0
1990 3-20-1 7-17-0 3-9-1 7-4-0
TOTAL 114-358-7 123-349-7 114-99-4 123-129-4
</pre><hr />
Here's the numbers where spread >= (1/3)(total).
<font class="small">Code:</font><hr /><pre>Season Over/Fav Under/Dog Fav Dog
2007 1-1-0 1-1-0 1-0-0 1-0-0
2006 2-5-1 2-6-0 2-0-1 2-3-0
2005 1-6-0 3-4-0 1-1-0 3-2-0
2004 0-2-0 0-2-0 0-1-0 0-1-0
2003 0-3-0 0-3-0 0-2-0 0-1-0
2002 0-3-0 1-2-0 0-1-0 1-1-0
2001 2-3-0 0-3-2 2-0-0 0-1-1
2000 4-6-0 2-8-0 4-3-0 2-1-0
1999 2-6-0 4-4-0 2-2-0 4-0-0
1998 2-3-0 0-5-0 2-1-0 0-2-0
1997 1-5-1 2-5-0 1-3-0 2-0-0
1996 0-4-0 2-2-0 0-0-0 2-2-0
1995 1-7-0 3-5-0 1-1-0 3-3-0
1994 3-6-0 4-5-0 3-1-0 4-1-0
1993 4-10-0 6-8-0 4-2-0 6-2-0
1992 3-24-0 6-21-0 3-7-0 6-11-0
1991 3-11-0 5-9-0 3-4-0 5-2-0
1990 1-10-0 4-7-0 1-3-0 4-3-0
TOTAL 30-115-2 45-100-2 30-32-1 45-36-1
</pre><hr />
Firstly, it definitely looks like the spread being 1/4 of the total is not enough. If the probabilities of the events were 50% and there was no correlation (on average) you would expect about a 1:3 ratio of wins to losses. 237-707-14 is almost exactly that.
When the spread is 1/3 of the total, you start to see something. It looks like the under and the dog may have some correlation. Since even when the favorite covers, the game fails to go over roughly 50%, there does not seem to be much of a correlation between over and favorite here. Anyone want to chime in on statistical significance? Or is this like seeing a pattern where none exists?