[LMU-OSC News] Invitation: ReproducibiliTea - Session 9

[LMU Open Science Center News]

Hello all,

we would like to invite you to the ninth meeting of our Journal Club “ReproducibiliTea” which will be the last one for this semester. It will already take place this Friday from  2:30pm to 4:00pm. We are very happy to welcome Prof. Chris Donkin (LMU Munich) as our guest who will give the introductory talk and participate in the open discussion afterwards. This time the session will take place online via Zoom (see link below).

When? Friday, 29th of July 2022, 2:30pm-4:00pm
Where? via Zoom: Meeting ID: 917 8852 1090 / Password: Replicate (https://lmu-munich.zoom.us/j/91788521090?pwd=QmJPeEZLcGtYeGlmeGFnazVoUVFzQT09)
Paper: Szollosi, A., & Donkin, C. (2021). Arrested Theory Development: The Misguided Distinction Between Exploratory and Confirmatory Research. Perspectives on Psychological Science, 16(4), 717–724. https://doi.org/10.1177/1745691620966796
Procedure: Prof. Chris Donkin, one of the paper’s authors, will start with a short talk about the paper (roughly 20-30 min). We will then continue with an open discussion in which you can contribute your own thoughts and questions.

It would be helpful to have read the announced paper, but there is no obligation to do so. We prepared the following key questions to guide you through the discussion and also provide some guidance while reading the paper. For each question, you can think about the suggestions from the paper, but also bring in your own ideas:

1. How might the flexibility of theories contribute to replicability problems?
2. What are the characteristics of good theories?
3. Is preregistration misguided to prevent QRPs? What could be arguments for and against such an opinion?

Below, you can find a short abstract for the introductory talk by our guest:  Prof. Dr. Chris Donkin (LMU Munich)

Feel free to contact us if you have any questions and/or remarks. We would also appreciate it if you could forward this email to anyone who might also be interested in attending the session (e.g. colleagues from your study/working group).

We are looking forward to seeing you soon!

Your organisation team
Laura Goetz, Stephan Nuding, Leonhard Schramm


Better theory development as a response to the prevalence of QRPs
Concerns about so-called QRPs, and especially p-hacking, are typically based on concerns about false-positive statistical outcomes. However, the outcome of a statistical test is not equivalent to a scientific claim, and so elevated Type-I error rates do not speak directly to reduced replicability. We will explain why certain kinds of statistical analyses, including those forms of p-hacking that seem most egregious, will indeed lead to an inability to predict the outcome of future experiments. The problem, simply put, is that without sufficient scientific argument, any statistical analysis taken at face value is an ad hoc explanation, from which there is no good reason to expect to understand the world. We contrast such explanations with good theories and discuss what those are. Thinking about p-hacking from this perspective helps provide a way of deciding what kinds of analyses are good or bad, independent of whether they are p-hacked, ‘exploratory’, or ‘confirmatory’.

About the speaker
Chris Donkin is a Professor of Computational Modelling in Psychology at LMU Munich. His work focuses on explaining cognitive phenomena, such as memory and decision-making, often relying on using mathematical and computational models as aids to help understand and test our theories.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lrz.de/pipermail/lmu-osc/attachments/20220725/ad9c7edb/attachment.htm>