I think I can intuitively understand yards or plays per point, but it's different to see points in the denominator. What if you flipped it to points/yard gained or points/play? It should still show the same conclusion, but maybe in a more common or easily understandable format? 

Thanks for taking the time to gather all the data. How do these teams compare in terms of scoring drives? Do they score on most of the drives or only in a handful? For instance, a 2 play TD drive for 90 yards  would include a long play. A 16 play TD drive of 90 yards would achieve the same yds/point but different plays/point. Who is more efficient? Who kills more clock thus eliminating couple of chances for the opponent? 

It's a good thought and has the potential to be a good measuring stick/discussion point, but any real conclusions seem incomplete without adjusting for opponents or short fields IMO.

Take PSU for example. Against Iowa they had 7 points from a pick that gave them the ball on the 3. 3 were from a FG after a big kick return where PSU didn't get a first. A drive where their offense didn't pick up a first down helps them here because they were already in FG range.

Against Indiana they scored 33 points. 7 were from a kick return to the IU 5. 7 were from a kick return to the IU 36. 3 were after a punt fumble when PSU only got 24 yards. So in reality that's 16 points from their offense on full field drives. Those two games alone hugely skew the numbers. And it's not even like the KJ 93 yard TD against OSU or the Sanders 78 run vs MSU because they made those plays happens. These are just drives where the PSU offense did nothing really but got gifted points.

does total points include defensive scores? if so, should take those out

Good analysis.  A couple methodological points:

1. The "more correct" way to do this is to have a historical database where you have information "the average team that started a drive from their own 32 yard line scored 2.9 points."  Then be able to compare individual drives against that.  As you said, short fields can skew things dramatically.

2. As I understand it, this is, to a large extent, a part of the FEI methodology.  S&P+ looks at things a bit differently (play-by-play), but FEI is more drive-by-drive in its analysis.

3. Obviously, junk drives/time should be excluded (or weighted lower) too.  When Alabama leads 77-3 in the mid-4th Quarter, their 0-point, 39-yard drive doesn't really matter much.

What about our defense? It feels like we give up a lot of points per yard.

Hasn't been like it the last couple of games, but last year and the first half of this year, it felt like we gave up all of yards on TD drives. There weren't many drives where we gave up a couple first down and then punt.

I like this and see two potential ways to account for the distortion of short, successful drives caused by returns or turnovers. The first would be to weight drives and points based on length. However, I suspect that would change the purpose of the metric. Alternately, you could simply filter out all drives that start inside the opponent's 30 (or 25 or 20) yard line. At some range, points should be assumed, either from a field goal or a touchdown.

Yes, points and yards per drive are the best ways to examine this.

Michigan intentionally plays at a turtle's pace on offense.

Using these same teams, I would like to see something similar for defense. Something tells me Oklahoma won't be near the top on that one.

Ohio Staters are griping about their historically bad 2018 offense.  Doesn't look too bad on paper.

1) I am not getting the exact same data as shown here - is your data is based on 7 games rather than 8? (I think the ncaa site may have updated shortly after you posted.) It's a little weird with only 7 games since the UM-MSU game would be included for MSU, but not for UM. 

2) I would be interested to see the data sorted a few different ways (e.g. by total points, or by total yards, or by yards per play):

Just eyeballing the numbers, it seems like yards-per-play is the most directly correlated to scoring output, while factors like total # of offensive-plays seem to have little or no correlation to scoring output (Indiana leads all teams in number of offensive plays by a significant margin, but is only average in total scoring).

Yards/Point seems to be somewhat (inversely-) correlated with total scoring, but the correlation seems much weaker than Yards/Play - Purdue is the third highest scoring team in the league but its Yards/Point is almost the same as Northwestern and MSU, which are the two lowest scoring teams in the conference not named Rutger.

Overall it certainly seems to the suggest that the "bad at turning yards into points" is definitely incorrect vis-a-vis Michigan, and more generally, that points-per-yard is probably not a good predictor of total offense for any team. The better route for putting more points on the board would seem to be to just focus on putting up more total yards/yard per play and that the points will tend to take care of themselves.

I've found the media's writing off our offense very problematic. We've very clearly placed an emphasis on the ground game (and TOP) with the intention of wearing down defenses over the course of the game, even if it holds us back from putting up big points. There is significant risk in this strategy, especially against a team like PSU. But at the same time, it has allowed us, at least in theory, to keep significant aspects of our offense under wraps as we come up to this crucial final run. I suppose we might want to test a more wide-open style under game conditions. But, on the other hand, if we are capable of that more wide-open style, we now have PSU, OSU, and any potential later opponents primed to play a team that plays a grind-it-out-on-the-ground style of offense, when it's far from inconceivable that we could (especially with Black back) open a game with an aerial attack that would completely knock a quality opponent for a loop.

Am I dreaming? Quite possibly. But the potential is definitely there.