Beyond significant

Ellis, Andrew (4 February 2019). Beyond significant (Unpublished). In: IRC (Interfaculty Research Cooperations) retreat. Institute of Psychology, Bern. 04.02.2019.

There has recently been much debate about the merits of null hypothesis significance testing (NHST). The American Statistical Association (ASA) recently advised researchers to avoid p-values, as these are commonly misused and misinterpreted. Bayesian inference is offered as an alternative; however, most researchers have received little or no training beyond the practise of NHST. In this talk, I will give a brief historical overview and a practical guide to current Bayesian data analysis methods. Amongst the topics covered are setting up a Bayesian model, parameter estimation, Bayes factors, model comparison and posterior predictive checking. The purpose of this talk will therefore be to familiarize the audience with Bayesian thinking and to demonstrate that with current software, Bayesian analysis is no longer restricted to a small number of expert statisticians.

Item Type:

Conference or Workshop Item (Speech)

Division/Institute:

07 Faculty of Human Sciences > Institute of Psychology > Cognitive Psychology, Perception and Methodology

Graduate School:

Swiss Graduate School for Cognition, Learning and Memory (SGS-CLM)

UniBE Contributor:

Ellis, Andrew

Subjects:

100 Philosophy > 150 Psychology
300 Social sciences, sociology & anthropology > 310 Statistics

Language:

English

Submitter:

Andrew William Ellis

Date Deposited:

06 May 2019 17:08

Last Modified:

06 May 2019 17:08

URI:

https://boris.unibe.ch/id/eprint/127670

Actions (login required)

Edit item Edit item
Provide Feedback