Mirror of the same bluesmoon / lognormal blog post on analysing performance data.
Blog post from bluesmoon / lognormal on how to analyse performance data. Includes some great points on geometric means and geometric standard deviation which better apply to lognormal distributions than arithmetic means / standard deviations. Very useful.
Really handy post with links to sources of web performance optimisation statistics.
Useful online book on statistics. A work in progress, but looks very valuable.
A good introduction to the statistics math of split testing.
Includes the math behind the calculation, and sample PHP code to actually generate the numbers. Very cool.
Simple formula to calculate statistical significance at 95% on AB / split tests. N = total clicks, D = diff between win & lose, if D² > N, it's statistically significant. Example, 14 clicks A, 22 clicks B, N=36, D=22-14=8, 8²=64>36, win. If 2D²>N, 99%.
QuantCast stats on the WordPress.com hosted network