@Topical_Island As the only person in this conversation with an advanced degree and years of experience in the field of statistics, the word is thrown around too much. Giving a single percentage, i.e. 60%, is what we call a point estimate. It is exactly that a single point in time and space. It has very little meaning without context added to it, and even less without considering the sample size from which it was derived. Smaller samples, generally mean larger confidence intervals and less statistical significance. In this case using ztest of proportions and null hypothesis of p=.5 returns, pvalues of 0.057 and 0.126 for Gush vs Shops and Gush vs Eldrazi respectively. Both of which would be statistically insignificant at a the standard alpha level of 0.05.
The sample size is simply too small. So lets hold off on demeaning the word of statistically unless we actually run these tests and they actually come out significant. In general though as I am not a frequentist, even I personally avoid using the word statistically if I can.
My figure of Gush being 30-40% was pulled from mtgtop8 and mtggoldfish, both of which are comparable in terms of collection as historical top 8 data. This is not a made up figure go for yourself and add up all of the decks playing Gush on those websites.
My statement about decks altering themselves to beat Gush? I was deriving that based on my experience, and the discussion here, primarily actually made by Mr. Menendian.
In my opinion, the data here says there are 3 decks. I was simply interpreting the results as they are laid out. The data could certainly be interpreted differently with different aggregations, different breakpoints of what you consider viable, ect. As much as other statisticians and scientists would have you believe there is much subjectivity in data analyses.