You are here

A Machine for Generating Exaggerations

Writing for Nautilus, Tom Siegfried disillusions readers about statistical significance: Just because a researcher reports a low P value doesn't mean that the effect in question actually exists.

Siegfried quotes an epidemiologist speculating that "false finds may be the majority or even the vast majority of published research claims"; reminds readers about the winner's curse; and enumerates the reasons why the studies journalists publicize are precisely those most likely to be statistical flukes.

Then Siegfried wraps it all up with a kicker about a fish.

Read the essay.

Start Date: 
Thursday, September 19, 2013