Lux Fortuna
*Anomalies: After the fact I realized that our 10.8 player did not have a major luck boost. This does not contradict the data, though, because she in fact received the -most- drops. The only way this would negatively affect the data would be if there is an inherent glitch in the luck system and there is no reason to believe that this is the case.
If we know the Luck of the players involved, it wouldn't be impossible for us to adjust the 10.8 value by the percentile difference. As you point out, though, it being the top data point makes this a little unnecessary.
Quote:
*Weaknesses of the Study:
-Very small sample size
-Short period of time
-Possibility that varying cl's within a group affect orb drop-rates (however unlikely the possibility for this is).
Sample sizes will always be a concern with these types of tests; the more information available, the better. Still, assuming around a 60% rate of orb drops, naturally, this test would only be about a hundred data points for each CL value. Unfortunately, I know DMS expects quite a large time investment for even that, so one can't really complain.
The one big concern in that respect, though, is that all of these results are roughly within the same value range - in limited-sample experiments, we tend to rely on very large distinctions from which to draw conclusions. In this set, the total range is only 12 - altogether, that's about 20% of our maximum value, which could reasonably be explained through the randomness of the system. Elsewhere in the game, overleveled players receive only half their normal drops, and those are the kinds of differentials I would be hoping to see in a small-scale experiment.
Quote:
Conclusions:
While I realize that this is a small sample size over a short duration of time the data can still be open to offer guiding advice.
-While the cl 10.8 received significantly more orbs than anyone else in the crew, it was expected that the 10.9 would exhibit similar results, however she suffered from 10% less drops than the cl 10.8 and her orb drops were practically identical to the 11.0, 11.1, and 11.2. I mostly attribute this to the small sample size and short duration, though.
-There seemed to be no significant difference between cl's 10.9-11.2.
-The cl 11.4 received the least amount of drops but this could have been a result of the short period of time. More research would be necessary in order to reach a stronger conclusion.
From the data at-hand, there appears to be a downward correlation between CL and orb drops; however, given the small differences in the middle of the data range, those values could just as easily have supported an upward or neutral correlation, had the two endpoint values change significantly - and given the random nature of the study, such changes can't be entirely ruled-out. As you point out, it would take more research to strongly determine where those endpoints should lie.
I may simply not be recalling it, but did you have a hypothesis for how you expected drops to function, before performing the experiment? I've heard before that 10.8 was considered the 'sweet spot', and that drops above it were reduced, but I've never heard by how much or whether the distinction was boolean (affects players above 10.8 equally) or scalar (affecting CL 11.4 players more severely, as the data tentatively suggests). Would you mind illucidating on how you suspect the system works, so we have a better idea of how the data is supporting or defying expectations?
250
50
200
