I don't view the game as an aggregation of statistical categories. I don't think it reflects the dynamics on the court.
Let's reset for a minute.
Randle and Horford played completely different roles last year, so they were not going to compete statistically anyway.
Randle is more talented, and more productive. Ok. But he's only more productive because he has a higher usage. The higher the usage, the more you're dictating the outcome of your team's possessions. And if you do it on low (abject in Randle's case) efficiency, the more inefficient your team's offense will become. Right?
Randle was horrible in his role last year. Horford was good in his, albeit a lesser role. But since Randle probably can't take on this high-usage role, can he take on a lesser role, and will he perform as well as Horford in that role? I personally doubt it.
Statistics like points and assists don't matter as much if you can't be efficient with a high usage because they will decrease with a reduction of your role. JVG mentioned Larry Hugues earlier and he went from averaging 22 ppg and 4.7 apg in Washington to 15.5 and 3.6 in Cleveland (on lower usage). That's my view.