The newly formed RSS Special Interest Group (SIG) in Teaching Statistics held its first session at the 2019 RSS conference in Belfast.
Simon Harden from the Department of Statistical Science at UCL chaired a panel discussion on 'Weaving statistical software through University teaching and learning'. The discussion kicked-off with the four panellists introducing their thoughts on the use of statistical software in teaching.
Elinor Jones from the Department of Statistical Science at UCL described the importance of visualisation in learning statistics and suggested that this might lead to R being the default package.
Meena Kotecha (pictured) from the Department of Statistics at LSE fully supported using software including Microsoft Excel, R and SPSS in teaching non-specialists, stressing the importance of theoretical understanding to facilitate effective interpretation and communication of statistical analysis.
Jamie Sergeant from the University of Manchester’s Centre for Biostatistics discussed his experiences teaching healthcare scientists and professionals, including using Jupyter notebooks. He questioned whether any of the claims made about the teaching of statistics were evidence based.
Bill Browne from the School of Education at University of Bristol described a wide range of uses for software in teaching and wondered whether packages such as Genstat or MLwiN might be more useful than R for some disciplines. He stressed that the practice of using software was as or more important than theoretical understanding for applied researchers.
A lively debate ensued, with the audience responding to the panellists’ initial thoughts and adding their own. Contributions from the floor represented a wide range of statistics educators: those working in higher education teaching both specialist and non-specialist students, those working in government, and those teaching short courses mostly to non-academics.
One discussion covered the relevance of the data being worked on. Students of statistics who bring their own data to analyse are more likely to appreciate the statistical concepts needed and may be more motivated to learn suitable software. This isn’t always possible, but using real datasets relevant to those being taught is an improvement over using generic data. Data with a more general theme but clearly of relevance to society, such as social justice, may appeal to students regardless of their specialist area.
Another question asked if all statistical teaching should involve the use of software? Some thought 'yes', while some disagreed that it was necessarily for theoretical concepts. Others noted that it's tricky when users, such as nurses, have a limited amount of learning time and their software use will be irregular.
Another conversations looked at command line vs menu-driven systems vs spreadsheets. There were fans and critics of all three, but it was clear that imbuing students with confidence, both in understanding statistical concepts and in using software, was key.
Read the abstract for this session, Weaving statistical software through University teaching and learning.