Yes, greater sample size is always good, but when 100% of tests show the complete opposite of what you assert, we must either say that they are going to become the outliers and thus discarded (odds of that are actually rather astronomically low) or we must accept that they they are a valid data set. I have asked for examples from you and provided my own unbiased examples, look at my schedule I only tested the ones I mentioned with no cherry picking. So odds of strat center not being in effect with the data points provided are astronomically low. If i were able to place a bet on the topic, i would bet the house. Feel free to show examples that show that my data points are in fact outliers, i love this kind of info.