It sounds like from the description that the percussionist was playing to a click and not with other musicians … and if I’m not mistaken, they evaluated the percussionist's performance relative to the click.
When I record, I often play to a click without music and without fellow musicians (similar to the way I am picturing this experiment). My personal perception is that, in this situation, I am not the principal time keeper but the secondary time keeper responding to the click. If I feel myself deviate slightly from the click or maybe play a fill or accent pattern … I am always trying to come back to the click, which in my opinion, is not necessarily musical (I honestly try to become one with the click but sometimes I find myself battling it).
When I play live or record with a live band, I am (more often than not but, not always) the primary timekeeper. And being human, there are fluctuations in my playing. The difference is that I am not trying to get back to the click but, the other musicians are responding to my playing and I to them. The result is that there is an inherent ebb and flow … albeit, tight rhythmically with solid time.
I know what I am about to say is anecdotal and not based on statistical analysis but, after looking at the transients associated with my drumming (in Cakewalk) for the last 15 years or so, there is a difference in the way I play when I am the primary vs. the secondary timekeeper.
I won’t attempt to extrapolate that statement to all drummers but …
I will venture that using statistical data to determine how a percussionist’s playing deviates from a click and then in turn use those results to humanize a drum program may ultimately get you the results you are looking for in a panel study (I suspect that high profile pop music is evaluated in corporate panel studies before money is used to back it … again, just speculation) but to call it humanizing is a misnomer.