13 bookmarks. First posted by FradSer november 2017.
> Power analysis of historical gwern.net traffic data demonstrates that the high autocorrelation yields very low statistical power with standard tests & regressions but acceptable power with ARIMA models. I design a long-term time-series approach in which a A/B-test running January-October 2017 in 2-day blocks of ads/no-ads uses client-local JS to determine whether to load & display ads. The final results yields a strong estimate of 14% traffic loss if all traffic were exposed to ads (95% credible interval: -13-16%) and an expected traffic loss of ~9.7%, far exceeding the decision threshold for disabling ads and rendering profitless further experimentation. Thus, banner ads on gwern.net appear to be quite harmful and AdSense has been removed.gwern statistics
november 2017 by porejide