Skip to content

To p(-value) or not to p(-value): Is that the question (with apologies to The Bard)

This CAN event took place on May 14, 2015.

Focus Theme: Talks and discussion on statistical measures of evidence and scientific learning

Event Description: The appropriate role of p-values as a measure of evidence against a hypothesis, and more broadly, null hypothesis significance testing (NHST) have long been subjects of debate.  Posterior probabilities of hypotheses and Bayes factors have been proposed as alternative measures of evidence.  Earlier this year, the editors of the journal "Basic and Applied Social Psychology" announced a ban on p-values in that journal.  Statistical societies have weighed in on the ban.  George Washington University faculty members George Howe (Psychology), Refik Soyer (Decision Sciences), and Sudip Bose (Statistics) will give short talks on the topic, followed by attendees' comments and discussion.  Topics will include the uses and limitations of NHST in causal modeling and the social sciences; Bayesian alternatives to the p-value, and the role of the likelihood, and ideas about agreement between frequentist and Bayesian hypothesis testing.

Please join us for what promises to be a lively discussion on the important and timely topic of statistical measures of evidence and scientific learning.
This event is organized by Professor Sudip Bose of the Statistics Department.
Send your suggestions for future events to ccasres@gwu.edu