This workshop is intended as a general introduction to using statistics in experimental settings, and to using the R statistics package. It will cover: getting data into R; descriptive statistics; inferential statistics; some standard parametric and non-parametric tests; ANOVA and related methods; experimental design; bootstrapping; alternative Bayesian methods.
It is not intended to be comprehensive, and in particular, on this website we will try to collect further sources, such as online tutorials, relevant articles, etc. that you can use to extend your knowledge. As such, please help us by giving feedback on linked material or new suggestions for useful links.
Note this is the first time we have run this workshop so we may need to be adjusting things as it runs. We are also new to using R so may not always know the answer (but will try to find out!).
We've aimed at a fairly basic level: covering the sort of statistical testing normally taught to psychology and biology students; but assuming some computational and mathematical competence. For some of you, some parts may be obvious, so feel free to skip. Conversely, we may sometimes have assumed knowledge you don't have, or used unfamiliar terminology (there is a lot of jargon), so don't be embarassed to ask us if something is unclear. Please also work interactively - ask your colleagues to help - or work in pairs if you want.
It would be good if you can bring some of your own data along to the sessions, and try to apply the methods. We will discuss issues that arise particularly in the final session. Alternatively, if you are reading a paper that has an analysis that you don't understand, bring it along for discussion.
In preparation we have drawn on the textbook Statistics: An Introduction using R by M.J.Crawley, amongst other sources. You might decide it is useful to purchase (we'll have some copies at the session) but let us know if you have other recommendations.
Monday Dec 9 2-4pm
If you are familiar with Matlab, you may find this Matlab/R reference useful.
Alternatively if you are familiar with Python, there is some helpful information here.
Tuesday Dec 10 2-4pm
You should look at this recent paper on problems with low-powered studies in neuroscience: Button et al. (2013).
Wednesday Dec 11 2-4pm
Thursday Dec 12 2-4pm
The potential for misuse of covariate analysis to obtain essentially any result the experimenter wants is discussed in Simmons et al. (2011).
Incorrect analysis of interaction designs is discussed in Niewenhuis et al. (2011) , who find that around half of neuroscience papers in top journals commit a basic error.
The problem of pseudoreplication due to incorrect analysis of repeated measures designs in recent neuroscience papers is discussed in Lazic (2010).
Friday Dec 13 2-4pm
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh