News for hackers. Stuff that matters.
Re:Most Winning A/B Test Results are Illusory [pdf] by tel on Sunday February 23, 2014 @08:51PM (#7288676)
A/A testing should be used to get accurate estimates for within-sample variance. If you run an A/A/B test then you can calibrate the A/B component to be sensitive w.r.t. the tolerances of real data.

And then yeah, I'm sure a lot of successful A/B tests will get washed.

There are no comments on this comment yet.