He had 2 drops. That is an extremely small sample size. Calculating percentages from 1 or 2 drops at this point in the season is worthless.
Beasley had 1 less drop but he also has less targets.
Yes, that's what the article says. So what does a lower or higher expected completion percentage based on route type have to do with anything, and what does a more accurate or less accurate QB have to do with anything?
Drops are only counted when passes are on target. Expected completion percentage based on route is irrelevant because such completion percentages are based on the need for a better throw when a player is further down the field than when he is 5 yards from the LOS. Because drops only occur on plays where the throw was good, it's hard to understand why we broaden our scope to plays where the throw was bad. The expected completion percentage of the route doesn't make a dropped pass any less on target. A dropped pass is inherently on target and the pass was good so it doesn't matter where it was thrown. It was catchable and yet it was not caught. QB accuracy is also irrelevant for the same reason. If we're only looking at the times where the QB delivered the ball on target, is there any reason to reference the times when he did not?
If you want to take it a step further, defensive play is also largely irrelevant because incompletions that are a result of the DB are not counted as drops. Those are tallied in the, "Pass Defended", column. Whether or not the defense was playing off or playing on doesn't matter because contested passes that fall incomplete are credited to the defense. Therefore, we're only look at the uncontested passes that fell to the ground.
Is there something else you meant by listing these very reasons for why a comparison of drops is inappropriate?
Dave Chappelle had a video/skit "When keeping it real goes wrong". I would like to change that to "When playing with stats goes wrong".
I imagine it would be a great episode, too. Probably star a guy who throws out 2 or 3 factors that are more or less non-influential to the discussion at hand and then proceeds to get all smug about it when someone points out that he might be looking at how the data is measured from the wrong perspective.