You are looking at blended statistics across games under the premise that every throw reflects a QB’s accuracy ability the same. The flaw with this approach is that it assumes (falsely) that every throw is made under the same conditions.
Thats why I suggest looking at how accurate the QB was in each game (I.e. how often on a given day, the QB threw accurately), regardless of how many attempts they had in the game - to mitigate the impact of outlier performances due to match up, situational need to force throws playing from behind, etc. It’s here where Gavin’s accuracy struggles really stand out comparatively because you observe that he only completed half of his passes 4 times on the entire season (one of which, Wagner, was basically a scrimmage). It didn’t matter what the situation - clearly, GW was consistently inaccurate with passing for whatever reason.
Our new QB, completed at least half his passes in 8 out of his 12 games, and completed over 57% of them in 6 games (half of them). The blended stats your looking at double weight games like Purdue and Nebraska (almost 30% of all his passes were made in those 2 games) where he didn’t even play bad but was forced to hoist 42+ passes because his team couldn’t esblish the run at all (Nebraska) or he needed to try to keep pace with the 49 points his D gave up (Purdue).