As a warning, I did not read all the comments for this post. I apologize in advance if some of my questions have already been answered or discussed. Most of my comments will be under the premise of this statement, for I would like to leverage your expertise/knowledge. Can you share your evidence for this (numbers would be very helpful, but a repeatable procedure is sufficient as well)? Did you conclude the "very clear" from observation (as opposed to hearing it from the devs, or some other source)? If so, how? Again, could you expand on how you concluded the "very clear" (if it was from observation)? If you have some historical data, can you make that available? Is "multiple dozens of completion records" sufficient to conclude something about EBs? If so, how did you determine that and how long will that stay true (my intuition tells me that as time goes on, the "multiple dozens" would have to grow to "multiple [hundreds|thousands|etc.]")? How many is "multiple dozens" and "numerous occasions"? How did you observe this in a way to conclude the "appear to have no statistical correlation"? The method I can think of would've taken quite some effort. Was there a trick to it? Did you automate something? ~Can you teach me how to verify these results in a statistically sound manner? It would also be nice to get your data as well so I can know that I'm not doing something wrong. You don't have to explain the specific statistical processes to me, just point me in the right direction. Hopefully I'll understand everything eventually. Thanks