Appalachian State defeated Michigan in the first game (for both teams) of the 2007 season. This loss (to a non Division 1-A program) had an effect on all the Rewards rankings generated (and listed on the previous web page). Non 1-A teams were 82-413-1 in the 1990's, and 65-445 from 2000 to 2007, and of those wins, only 11 were against teams whose final record had 2 more wins than losses (4 times, it was Nevada). Those wins were against the following teams: 1992 Nevada (7-5), 1993 Western Michigan (7-3-1), 1993 Nevada (7-4), 1994 UNLV 7-5, Nevada (9-2), 1996 Nevada (9-3), 2002 Northern Illinois (8-4), 2003 Navy (8-5), 2004 Hawaii (8-5) and North Texas (7-5) and finally Michigan (9-4) in 2007. Because App. State is lumped into the NON_DIV1A "team conglomerate" (whose record in 2007 was 9 wins, 71 losses), a loss to any non 1-A team garners an almost full point reduction (in the final Rewards ranking) for that team who suffered such a loss. If one were to apply the PRS (the Power Rating System, which is used by the Rewards system) to evaluate all 719 teams in Peter Wolfe's data file, excluding any bowl games, App State ends up rated as the #67 team by PRS, with the full scores, and as the #36 team when the margin of victory is reduced to 1 point, i.e. PRS_1, after all post season games were completed. (App State won the "1-AA championship", and finished the year at 13-2.) Therefore, this one game has a tremendous effect on the ratings: including it (with App State = NON_DIV1A) lowers Michigan's rating (#31 in PRS_1), thereby lowering Ohio State's Rewards ranking, and removing the game causes Michigan (and Ohio State) to be rated higher - #17. Michigan ranked as the 21st team overall, from the set of all 719 teams (in the PRS_1 modification), signifying that its "true" PRS_1 rating is closer to the one generated when ignoring this loss.

Ohio State ended the regular season as the #1 team in both polls, but by including Michigan's loss to App State as if it were to a NON_DIV1A team, Ohio State ends up as the #7 team (as determined by the Rewards system). Averaging the PRS_1 ratings when this game is included as a loss to NON_DIV1A with the PRS_1 ratings if the game is removed raises Ohio State up to the #5 position in the Reward rankings (before the bowl games). If the loss were changed to a victory for Michigan over NON_DIV1A, Ohio State would end up ranked as the #3 team, just behind Virginia Tech, and if the loss were totally removed/ignored, then Ohio State moves up to #2, and then the Rewards ranking system would agree (once again) with the two teams that were selected for the BCS National Championship game. (My guess is that if this loss were modified to have come from a reasonably good 1-A team, then Ohio State  would remain in the #2 or #3 position in the Rewards ranking. Unfortunately, this hypothesis can not be "validated" because the adding of such a game would affect all the other PRS_1 ratings, especially the team that is given said extra win as well as all of its opponents.