# (in)Significant Results View this submission here: http://e.pc.cd/Ol9otalK ## Blurb I have tried to demonstrate as many errors in a statistical study through a visual art form emphasizing absurdity. The original question of determining the color of sheep's wool based on finite samples demonstrates an inherent (sample dependent) problem with the Frequentist approach which is typically circumvented by having a good randomized design and not cherry-picking the data. We demonstrate also the interpretations from different disciplines of observations based on rigor, the mathematical or ultra-pedantic approach would assert that a single side of a white sheep might be painted, while Bayesian's would update the prior probability and frequentists, their samples / population assumptions. Here we show both cherry picking, and running the experiment "until significance" without any tests of power or effect size. Additionally, I demonstrate the effect of a poor understanding of the p-value and the media's lack of nuance when it comes to reporting statistical studies (correlation is not causation). In the end, we demonstrate how a Bayesian approach with a prior (say, that sheep wool can take a variety of colors and is location independent) could have also circumvented the issues demonstrated, by allowing a degree of belief to be quantified. This work was also inspired by two [XKCD](https://xkcd.com/about/) comics, numbers [882](https://xkcd.com/882/) and [1132](https://xkcd.com/1132/).