I respect the source as a fellow math guy, but these statistics appear to be flawed.
The most obvious incorrect result is the "Median = 0" for +11 to +12. The median must be greater than or equal to 1 by the nature of the problem (you can't go from +11 to +12 in 0 ore).
The results don't indicate whether they are for a user switching to normal Bradium/Carnium for the +10->+11 roll, but from the mean values for +11->+12 I think it can only be realistic if they are NOT switching to normal ores for the free roll.
I agree with getting rid of the median value from the calculations. Sample size is stated at 100k attempts, so I don't see how skewed sample size can be the issue. But I also agree it's not terribly useful and wouldn't spend time on trying to adjust it.