I'm too lazy to look this up, but could the offensive numbers be skewed by an out-of-the-ordinary number of instances when the team is down one or two scores in the 4th quarter? When a team is down one or two scores in the 4th quarter, the opponent often goes into a "prevent" type of defense that gives up short plays to prevent the big plays. If you get the ball 2 or 3 times in the 4th quarter under those circumstances, you can count on 50 to 100 "easy" yards that likely wouldn't have been gained if the opponent remained aggressive on defense.
In other words, an average offense can post above-average numbers if it has significantly more opportunities against a "prevent" defense.
On the flip side, defensive numbers could also be skewed a bit if a defense never has to play "prevent" defense and the opponent's offense goes "conservative" to avoid an interception.
Sooooo, if my totally unscientific theory is correct, a team can have deceptively good numbers on both sides of the ball if it plays a large number of close games that it loses.