Research showcase: Too big to be true: science without exaggeration
Date: Mon, 17 September 3:00pm - 4:00pm
Location: ECU Joondalup, Room 1.447
Presenter: Professor Tom Stanley
RSVP to this event: Register here
Professor Stanley has applied meta-regression analysis in a systematic review of scientific results. This event showcases his results and illustrates and explains the 'decline effect' of scientific facts over time.
For a quarter of a century, Professor Stanley has been developing and applying the statistical methods of meta-regression analysis to objectively summarize and evaluate empirical science. Meta-regression analysis is a type of systematic review that summarizes and statistically explains all the far-ranging scientific results on a given subject. In the process of this research, much has been revealed about specific economic phenomena and policy questions, and several trends have emerged about economics and science. In particular, scientific ‘facts’ are typically much smaller and less consequential than commonly believed and reported. Comprehensive and rigorous statistical analyses of individual areas of scientific inquiry often find that the effect in question is much smaller than what the research literature typically reports.
In a recent New Yorker article that has receive a lot of attention, “The Truth Wears Out,” Jonah Lehrer documents a general tendency for all scientific facts to get smaller over time—the ‘decline effect.’ However, this effect has long been known by meta-analysts, and Prof. Stanley has spent a decade developing methods to detect and correct it.
Prof. Stanley’s talk will illustrate and explain this ‘decline effect’ using examples from medicine and economics. It will also discuss how methods to correct for publication selection bias (or reporting bias) offer a ‘Science without Exaggeration.’