Reminder of today’s seminar
As part of our Behaviour and Evolution Research Group (BERG) seminars, Wednesday (19th October, 5.30pm ) we have Professor
Vernon Gayle, University of Edinburgh giving a short presentation on:
Reproducible Research is ‘Show Me’ not ‘Trust Me’
Psychology Common Room, 3A94, Cottrell building, with usual drinks and nibbles.
I think this topic is really important to consider in terms of our research process and workflow.
Hope to see some of you there.
Best, Hannah
Abstract
Despite the explosion in the number of research publications it is impossible to ‘reproduce’ the results of most analyses that are published. This is because information
on how the work is undertaken in seldom made available. Most researchers will have a happy, or possibly even a terrifying, early educational memory of being told to “show their working out”. Somewhere between primary school and graduate school this requirement
has evaporated. Currently published research is far from transparent and a culture of 'trust me’ rather than a culture of 'show me’ exists.
In this presentation I make an appeal for researchers to routinely provide enough information so that others can check that results are accurate, and that correct inferences
and conclusions are reported in published work. This transparency will also allow others to test the robustness of the original piece of research, for example by employing new or additional data and alternative methods. I highlight the obstacles that are
commonly encountered and provide some practicable steps for rendering research ‘reproducible’.
I argue that conventional research publications should, at best, be regarded as a palimpsest of the ‘real work’ undertaken within the data analytical process. Drawing on
insights from computer science and other disciplines that have been engaged in e-Research I illustrate how contemporary digital resources could provide a useful and effective aid to making social science research more easily reproducible. I propose a set of
guidelines which researchers should follow in order to enhance the reproducibility of their research. I conclude by suggesting a set of benchmarks against which the reproducibility can be assessed.
Hannah M. Buchanan-Smith
Professor, Behaviour and Evolution Research Group (BERG)
Psychology, Faculty of Natural Sciences
University of Stirling
Stirling, FK9 4LA
Scotland
Tel: 01786 467674
Fax: 01786 467641
E-mail: h.m.buchanan-smith@stir.ac.uk
Home page: https://rms.stir.ac.uk/converis-stirling/person/11925
http://www.247animalwelfare.eu/index.html
Just launched: http://refiningdogcare.com/