Conférence : Clinical psychometrics: thinking sensibly in our 'post-p-values' world

Publié le 24 février 2020 Mis à jour le 30 août 2022
le 9 mars 2020
Maison de la recherche
en salle D31
Maison De la Recherche - salle D31

Le professeur invité du CERPPS/Pôle 1, Pr. Chris Evans

The ASA 2016 statement about p-values summarised arguments within statistics that go back nearly a century but link with huge concerns about the "replication crisis" or "reproducibility crisis" which is very real in psychological research.  Some, not all, of the crisis does come from a long tradition, in all fields that use statistics, of totemising and fetishising statistical "tests" and "p-values".  The problem is just as serious in psychometrics though it's less about p-values there.  In this seminar I'll say a little about my 30 years of work with clinical (and some less
clinical) questionnaires. That has included work on the Body Shape Questionnaire (www.psyctc.org/psyctc/psyctc-org-home/instruments/body-shape-questionnaire-bsq/),
the CORE system (www.coresystemtrust.org.uk/), PSYCHLOPS
(http://www.psychlops.org.uk/) and, most recently, the RDFS (www.researchgate.net/publication/331876602_Relational_Depth_Frequency_Scale).
I'll argue that we need to change our thinking about psychometrics, in line with how we will be changing our thinking about p-values, to move from dangerous madness which Spiegelhalter should become actionable research malpractice (Matthews, Wasserstein & Spiegelhalter, 2017), to "psychometrics as principled argument" to paraphrase Abelson (about statistics) and to New Zealand PPDAC: Problem, Plan, Data, Analysis, Conclusion & communication approach to statistics (MacKay & Oldford, 2000).