Interdisciplinary learning doesn’t just apply to university students. In this post, Dr Jill MacKay explains how she learnt about a colleague’s innovative teaching tool in the School of Engineering, and, with his help and expertise, applied it to her teaching in the Royal (Dick) School of Veterinary Teaching, to help students better understand the nuances of statistics…
When I saw Alan Murray talk about the ‘toybox’ tools at the University’s first Learning and Teaching Conference, it was the playfulness of the resource that really appealed to me. To the uninitiated, a spreadsheet might not seem very fun, but the toolbox tools that Alan showed off were all about pushing relationships between variables to their breaking point.
Testing relationships, rules and boundaries is exactly why play can be a useful tool for the educator. When we play, we pick a set of arbitrary rules to follow, and discard rules that aren’t useful to us. Play is a great equaliser, and is commonly said to allow people (and animals !) the opportunity to explore new roles in a safe setting.
I teach data handling to both postgraduate and undergraduate students at the R(D)SVS and, for a couple of years now, I’ve been using fictional datasets to encourage students to get stuck in and play with data earlier. There’s nothing sacred about data that means you can only test ‘real’ numbers! For example, I have conducted surveys to dragons about their motivations for kidnapping princesses, and run experiments on unicorn farms to improve their magic dust yield. I think this is particularly important for our distance learning students, who sometimes perceive that the on-campus experience is superior, with greater contact with lecturers, particularly when doing their research dissertations . In my course, I want students to continue the transition into independent practitioners, and I’ve found that this sort of change really benefits from light-hearted examples.
The first toybox that I’m going to use for my MSc students next year comes from the aforementioned unicorn dataset, and demonstrates the relationship between the level of ‘noise’ in a dataset, and how this changes the ‘statistical significance’ of the data.
With Alan’s help, we developed the unicorn magic dust dataset toybox in Excel. As you can see in Figures 1 and 2 below, by moving the ‘noise’ slider, the student can explore how the F value, and associated F distribution (or P Value) changes. While these are both ‘significant’ results, the noisier the data gets, the less confident we become in the relationship. By visualising the data at the same time, you can see why the smaller F value makes us a lot less confident in our claim that unicorns yield more dust when listening to gentle conversation. Students can play with the sliders to produce a dataset which still has a significant P Value, but maybe isn’t very meaningful.
I hope that students will take this toy and start pushing it to its limits. I want them to realise that these ‘hallowed’ P Numbers are just a description of the data, and that you need to interpret them carefully. The toybox is one more tool that we can use to help students visualise the relationship between data and statistics, and, if they can draw conclusions about unicorn farming, they can certainly draw conclusions about their own data.