Ahh, this helps me understand the data and percentages a little bit better. Out of 135k users, almost 120k of them were Member or less. I think this would skew the data a bit, because without equal representation within the data, or at least close to it, you cannot expect a good sample for comparisons with the higher rank. I know it might make the calculations seem more realistic if you include 135,000 users instead of 10,000, but the comparisons that would be drawn from equal populations within each would seem more representative.

Although, I don't know how representative these percentages of each rank are for the entire forum, so it may more or less be negligible, but for me it might be more helpful to have equal numbers of each rank when doing the data to get a more uniform representation for comparisons and conclusions.