Chris Tosh | The piranha problem in statistics

<p>The piranha problem (too many large, independent effect sizes influence the same outcome) has received some attention on Andrew Gelman’s blog. But now it’s a paper!  Chris Tosh (Memorial Sloan Kettering) talks about multiple views of the piranha problem and detecting the implausible scientific claims that are published. The butterfly effect makes an appearance. </p> <p>If you enjoyed the science-vs-pseudoscience topics, you’ll enjoy this one.</p> <p> </p> <p>0:00 - Coming up in the episode</p> <p>2:35 - What is the Piranha Problem?</p> <p>19:54 - Confusing effect sizes</p> <p>23:11 - The "words & walking speed" study</p> <p>26:22 - Declaration of independent variables</p> <p>30:58 - Piranha theorems for correlations</p> <p>37:07 - Piranha theorems for linear regression</p> <p>40:37 - Piranha Theorems for mutual information </p> <p>44:13 - Bounds on the independence of the covariates</p> <p>46:12 - Applying the piranha theorem to real data</p> <p>50:12 - Applying the piranha theorem across studies</p> <p>54:05 - A Bayesian detour</p> <p>1:00:12 - The butterfly effect & chaos</p> <p>1:04:26 - Applying the piranha theorem to cancer research</p>