Odds Are, It's Wrong: Why Newsworthy Research Is Often Erroneous

Saturday, 14 February 2015: 3:00 PM-4:30 PM
Room LL21D (San Jose Convention Center)
Tom Siegfried, Science journalist, Springfield, VA
Science journalists are told they should always make sure that the scientific results they write about are statistically significant, in addition to applying other traditional criteria of newsworthiness. In recent years, a number of statisticians and other researchers have called attention to serious flaws in the interpretation of experimental data, especially the use of statistical tests of null hypotheses using p values. As a consequence, many experimental studies draw erroneous conclusions from their data. Such erroneous studies are commonly those that offer the first report of a finding (the “winner’s curse”), those in hot research fields, and those that report a finding “contrary to previous belief.” Diabolically, being first, being in a hot field and being contrary to previous belief are three of journalism’s prime criteria for newsworthiness, leading to news coverage that is inevitably skewed toward experimental findings that are likely to be incorrect.