Even putting that aside, the change in expected point value

is big. Not that many single plays in a game carry a -3 change in expected point value.

Yes, a dropped pass in the end zone is a -6 change in expected value. And in the case of Jeffrey's a dropped pass in the red zone is probably greater than a -3 in expected point value.

Likewise, giving up a big play by a DB can have the same effect or missing an open receiver for a TD (a -6 in possible value change), but replacement level coverage will give those up over the course of the game and even a good qb frequently misses those receivers. So another thing to consider is the degree of difficulty of the error. A possible point value change isn't the same thing as an expected value change. If an average qb (or this particular qb, depending on what you want to look at) only hits that throw 50 percent of the time, then the expected point value of the play is 3 points and the failure to execute is a -3 change in value.

On the other hand, it is more rare to drop Jeffrey's ball than it is for a receiver to find a soft spot in the zone for a big gain. So Jeffrey's drop is much more blame worth in identifying the cause of the loss.

Of course, underlying all that is that we have no idea if the rest of the game would play out the same way if you change an earlier event. Which is where the likelihood of winning change you cite comes into play.