Wednesday, May 04, 2011

Introducing StatsBytes - Introduction to statistics with R

R I am happy to announce the launch of StatsBytes  - a free, online self-directed introduction to the statistics package R. The documents as published today are essentially a first draft and I hope everyone who read this will help by scanning them carefully for mistakes, which I am sure are lurking there. Teaching materials of this sort need to be embedded in academic disciplines for acceptance. StatsBytes is published under a CC-BY-SA licence, so I want people to take it and repurpose it by rewriting the examples. Once the content settles down a bit, I may make a downloadable archive available to make it easier for people to do just that.

Originally, I planned to produce this resource as a microchunked uncourse on In practice, I found that this approach didn't help me with authoring in this instance (unlike MicrobiologyBytes, which has been my day to day notebook over the last few years and has contributed enormously to the production of the new edition of my textbook, Principles of Molecular Virology), so StatsBytes is now an old-fashioned static website. The one concession to modernity is a trial of a Facebook page as a support site for the course. I'm not yet sure if Facebook is a good choice for this or not, or whether better alternatives are available - I'd welcome your comments on this.

Unlike it's predecessor, StatsBytes does not (currently) include any screen capture videos, or indeed any appearances by Socky. This might change, but so far I feel that the text-based R interface does not really lend itself to video. Although I'm concerned about losing the impact video can have when it works, I'm not shoehorning video in there just for the sake of it.

I'm also thinking about my assessment strategy for my own students who will ultimately use this course. My current main priority is to reduce the over-assessment burden that students suffer in order to give them more time to engage with learning. Modularization is one of the sources of over-assessment, as is salami-slicing of continuous assessment, so bearing in mind the pass/fail competence test nature of our first year key skills module, my current plan is to make all the materials available with formative assessments which students can retake as many times as they want, and then to assess competence via a week long online exam in the penultimate week of term (to avoid clashes with other deadlines). I may need to slip in a midterm exam to relieve anxiety over failure to engage with formative assessment. To spend my time supporting increasing numbers of students rather than dealing with niggles over test results, the assessment format needs to be minimalistic and streamlined rather than complex and reiterative. Blackboard is currently giving me some grief in this regard.

I did think about trying to put a game layer on StatsBytes, but as ze Frank said at SXSWi, there's no point unless this is integral to the function of the site.  Anyone who is motivated enough (for whatever reason) to slog through a self-directed course on R doesn't need the added complication of a game layer!

Update:  For various complicated reasons, StatsBytes is no more! The plan is to produce a more static version in the near future.


  1. hurray! I'm assuming you still need me to plough through the rest of the course after my dismal failure in part one... personally I could have done with a video for the very first 'how to enter anything into a cursor type environment' part, but I take your point about it not suiting the system like it does with point and click.

  2. Re the old fashioned static website - might that not be a turn off for some? On a completely superficial level the page doesn't look particularly enticing and possibly not clear how to navigate. Have you thought about using a google site to tart (and chunk) it up a bit? Just a thought

  3. Anyone motivated to plough through this stuff is going to be pretty goal-directed, so fancy presentation may matter less - clarity is more important?