Uncommon Descent Serving The Intelligent Design Community

Pushback against abandoning “statistical significance” in science

What’s hot? What’s not?/Niklas Bildhauer, Wikimedia

Remember the recent call to abandon statistical significance in science research? There’s been some pushback:

Statistical significance sets a convenient obstacle to unfounded claims. In my view, removing the obstacle (V. Amrhein et al. Nature 567, 305–307; 2019) could promote bias. Irrefutable nonsense would rule.John P. A. Ioannidis, “Retiring statistical significance would give bias a free pass” at Nature

Ioannidis is a well-known scourge of bad data. See, for example, Another Well-Earned Jab At “Nutrition Science”

A statistician argues that two separate problems Are being conflated:

By focusing on the term ‘statistical significance’, we ignore the more important issue of what constitutes sufficient evidence of a true association. Let’s have that discussion and redefine what we mean by a statistically significant finding. Valen E. Johnson, “Raise the bar rather than retire significance” at Nature

From a trio of psychologists:

Without the restraint provided by testing, an estimation-only approach will lead to overfitting of research results, poor predictions and overconfident claims. Julia M. Haaf, Alexander Ly & Eric-Jan Wagenmakers, “Retire significance, but still test hypotheses” at Nature

Of course, as science embraces post-modernism, “irrefutable nonsense” could be the new standard. Along with ever more strenuous demands that we trust science.

Follow UD News at Twitter!

See also: Abandon statistical significance, learn to live with uncertainty, scientists demand Let’s see where this goes. Will it lead to less magic with numbers or more and bigger magic?

Um, are they suggesting that governments abandon statistical significance in testing the safety and effectiveness of new drugs? Or perhaps the apparent significance of high drug or alcohol levels in deadly automotive accidents? Or the apparent connection of sun spot activity to weather on Earth? I think the entire Insurance industry would collapse if they were required to stop basing premium rates for individual polices on decades (centuries?) of data linking the probability of future "accidents" to certain types of social behavior, neighborhood in which you live, level of education, etc., etc. But I'm guessing that they mean ONLY that some narrow range of research studies in Biology are immune to Normal Distribution and Pareto analysis. vmahuna

Leave a Reply