I have long been of the opinion that SEC defenses are overrated and that a large amount of their success is a result of the teams in the conference having little to no offensive imagination. Part of this thinking is based on the fact that an offensive guru like Steve Spurrier or David Cutcliffe or even Al Borges can basically run roughshod over the vaunted SEC defenses, but I did not know for sure so I went and looked at the box scores of the teams from 1995-2005, minus a handful of box scores that I could not find. Then I split up the data into non-conference and conference opponents because of my two hypotheses. In the data set, there were 471 non-conference games and 1,029 conference games
The first hypothesis was that if these defenses were so spectacular, then their defensive prowess would extend to their non-conference opponents and that they would perform even better defensively since these non-conference opponents would probably not be used to facing an SEC defense. The second hypothesis had to do with the offense and tested whether the SEC offenses exploded against non-conference opponents after leaving the suffocating SEC defense.
As for the first hypothesis, SEC defenses did perform better against non-conference opponents. They, on average, gave up 4.9 less points, 18 fewer passing yards, and 17.1 fewer rushing yards than when they faced fellow conference foes. Also, these defenses held their non-conference opponents to a lower completion percentage, yards per catch, yards per rush, and yards per passing attempt.
The offenses did also do better against non-conference teams. In these games, the offenses scored on average 7.5 more points, passed for 23.4 more yards, and rushed for 27.6 more yards. As is to be expected, they completed a higher percentage of passes and had a higher yards per catch, yards per rush, and yards per pass attempt.
Now, the two hypotheses could have been laid to rest there, but most of the non-conference foes that any major conference schedules are absolute jokes and of course these teams could be expected to beat up on the weakest teams in the country. Therefore, I took out all the non-BCS non-conference foes. I included Notre Dame in the BCS teams and also applied the definition of BCS teams retroactively (i.e, Louisville is included). Doing so left me with 183 games.
After I did that, the difference was erased completely. The SEC defenses gave up 1.6 more average points against their BCS non-conference opponents than against their intra-conference adversaries. The defenses did allow a slightly lower completion percentage and on average gave up .6 fewer passing yards, but gave up 9.9 more rushing yards. Every other stat I looked at was basically the same.
Like the defenses, the offenses performed incrementally better in some categories and identically in other categories. On average, they scored .7 more points against BCS non-conference teams and passed for 20.9 more yards while rushing for 4.3 fewer yards.
The question now is how these numbers stack up to the other major conferences around the country, but since I only looked at one conference, I can only give you the numbers for this one conference, but SEC defenses are what they are when facing comparable foes, giving up around 24 points (+/- 13 points), allowing 209-210 passing yards (+/- 90 yards), and 139-149 rushing yards (+/- 78 yards) a game. The numbers in parentheses are one standard deviation away from the averages.
On a different note, I also ran the correlations to see which statistics points are mostly tied to, for all opponents regardless of BCS conference affiliation. When SEC teams are on offense against non-conference foes, the stats' correlation are in this order: yards per passing attempt, yards per rush, rushing yards, completion percentage, passing yards, yards per catch, pass completions, rushing attempts, and pass attempts.
When SEC teams play offense against conference opponents, the correlations go like this: yards per passing attempt, rushing yards, passing yards, yards per rush, completion percentage, yards per catch, rushing attempts, pass completions, and pass attempts.
For the games for SEC defenses against non-conference teams, the opposing offenses points are correlated in this order: yards per passing attempt, passing yards, rushing yards, yards per catch, yards per rush, completion percentage, rushing attempts, pass completions, and passing attempts.
When SEC defenses play intra-conference rivals, the correlations are as follows: yards per passing attempt, rushing yards, passing yards, yards per rush, completion percentage, yards per catch, rushing attempts, pass completions, and pass attempts.
I never really answered my original question of whether SEC defenses are overrated, but I no longer think the defenses are a result of no offensive imagination within the conference since the defenses perform equally well in and out of conference unless their foes have no offensive imagination either.