In the end, the results determine the correctness of the decision. It is a difficult metric to be judged by, but this is why coaches make millions and why they frequently get fired.
1. Make FG, win game--decision correct.
2. Make FG and lose game by 1. Decision incorrect.
3. Miss FG or FG blocked. Decision incorrect.
4. Go for TD, unsuccessful. Miami scores on 99 yard run or pass or drive (Recall Syracuse had a 94 yard TD recently against us). Decision incorrect.
5. Go for TD, successful. Decision correct.
6. Go for TD, unsuccessful. Get defensive stop and punt. Kick game winning FG from 48 yards and win. Decision correct.
7. Go for TD, unsuccessful. Get defensive stop and punt. Miss potential game winning FG from 48 yards and lose. Decision incorrect.
There are many other permutations of this. Point is that there is a lot of football still to be played at that juncture and a lot of different things can happen. Overall success metrics are taken over a large sample and not corrected for confounding factors through regression analysis, so individual results may vary. It comes down to what the likelihood of scoring the TD is for that particular team in that particular situation, and what is the likelihood of the defense getting a stop. The fact that they didn't score on the first three downs influences the confidence in scoring on the 4th.
And applying NFL metrics to NCAA likely isn't valid. There is far less variation in skill levels between NFL teams/players than NCAA teams/players. A metric valid for Auburn vs Alabama isn't going to hold true for Rutgers vs Alabama.