Convincing other colleagues to use R in the classroom

Hello Everyone,

So, what I would like to hear from some of you at other institutions is what
can I do to convince/encourage my colleagues in other departments to adopt
R and save our institution a significant amount of money in licensing?

And 
Christopher W. Ryan noted:

"the key difference between point-click and writing code: reproducibility."

You may be interested in R software I have been using for 4 years now to support my second level undergraduate statistics for psychologists course. This allows instructors to develop R modules that permit interactive investigations of statistical methods online. Students' computations (including R code, data, results tables, figures...etc) are archived using a 'blogging' system, providing URLs they can embed in reproducible documents ('compendiums') written for their assignments. My course also uses an associated peer reviewing system enabling a socially constructive statistical learning environment. The same platform supports reproducible scientific writing and statistical analysis, and the developers of the platform now have a series of papers (I am a co-author on a few) you can consult about details at www.freestatistics.org . The latter hosts the reproducible computing platform and the statistical learning platform is at www.wessa.net/rfc Further details are available from the main developer Patrick Wessa who can be contacted via either route. There are scores of R modules already developed and freely available, and I believe course materials such as mine could be shared with instructors interested in developing their own online courses using the RFC platform too.

Patrick has talked at previous R-user conferences about this software, and I also gave a talk at the Warwick conference in 2011 on my experiences using it  - I think the podcast may still be available. Yes, they are here:
Teaching Statistics to Psychology Students using Reproducible Computing package RC and supporting Peer Review Framework
Abstract: http://web.warwick.ac.uk/statsdept/useR-2011/abstracts/310311-hollidayian.pdf
Slides :  http://web.warwick.ac.uk/statsdept/user-2011/TalkSlides/Contributed/16Aug_1115_FocusI_5-Teaching_1-Holliday.pdf

From the students' perspective they don't have to install (or pay for!) software, or learn GUI navigation or R programming and instead concentrate on learning about the statistical methods, and applying their understanding directly in writing their reports. From the instructor's perspective I have found it requires a little more support than a traditional course, but it is not very demanding. To develop your own R modules you need only be a standard R programmer like me - I'm not particularly adept with R. But as there are so many modules available - from t-tests and histograms to time series decomposition  and partial least squares path modelling I imagine most instructors will find their needs are covered.

There remains however a demand from my students (and colleagues) that students are exposed to SPSS as this is perceived as 'the gold standard' for statistical analysis, in psychology at least, and students want to be able to put 'proficient in SPSS' on their CVs. Note: not 'proficient in statistics'! Colleagues who have used SPSS for years (almost exclusively for anova and regression, tasks for which SPSS is, to my mind, a needlessly expensive solution) are unwilling to use other software and some even seem to find it difficult to recognize the same information when presented in an unfamiliar format. One even scribbled 'what's this nonsense?' on a student's work when sh/e chose to use QQplot and density plots from an R module in their research project! 

I also strongly support Christopher's comment about reproducibility potentially being a key driver for the uptake of R, but most scientific authors will not make their research reproducible until journal requirements or institutional directives force them to do so. These have both been tried and have failed, as papers still get published without meeting these requirements even when they have been in place. The one other potential driver for the uptake of R more widely would be powerful new statistical tools not available elsewhere that offer scientific writers increased publication success and ease of production. Here's where a platform like http://rfc.wessa.net/ possibly has a USP. Given an exemplar analysis one simply has to paste in new data into the data panel on an R module online and recompute the analysis allowing one to communicate new results within seconds without any local software or computation at all - you can even do this from your mobile device, as a lot of my students do.

Regards

Ian


Professor Ian E. Holliday PhD
The Wellcome Trust Laboratory for MEG Studies
Aston Brain Centre
The School of Life and Health Sciences
Aston University
Aston Triangle 
Birmingham B4 7ET
United Kingdom