Using Statistical E-books to teach undergraduate students quantitative methods and statistical software
Period of funding: April 2016 - April 2018
With the increased accessibility and volume of electronically available data, the ability to interpret, summarise and even analyse data has become a desired core skill for graduates in all disciplines. Within the social sciences there has been a push to increase and improve the levels of quantitative skills and the teaching of quantitative methods (QM) at all levels of university education. Initiatives like Q-Step for undergraduates, the advanced quantitative methods (AQM) pathway in the ESRC doctoral training schools and the British Academy’s quantitative skills programme highlight three of several schemes that have been put in place. Today quantitative skills are very much welded to the use of computer software to work with data. Although there is much merit in performing calculations or drawing graphs by hand to better understand the principles involved, in practice people will invariably use the computer to do the computations for them. It is therefore very important to also consider how best to incorporate statistical software in the instruction of quantitative methods.
At the Centre for Multilevel Modelling in Bristol we have a long history of computer software development. Our flagship package over the past 20 years has been the multilevel modelling software package, MLwiN (Rasbash et al., 2009). As statistical researchers our primary aim has been the development of software packages to allow the applied researcher access to cutting-edge statistical tools using appropriate methodology for the complex data structures that exist in real social science data. In recent years we have developed a second major software package, Stat-JR (Charlton et al. 2013) and here the emphasis of our work has changed slightly. In the time that we have been working on statistical software several things have happened. Firstly there are now more software packages out there offering the ability to fit complex models and second such software is used by not just the expert quantitative methods researchers but more generally by researchers in many disciplines and of varying levels of expertise. Stat-JR therefore, while still offering access to cutting-edge techniques and developments, also offers interoperability with the many software packages that exist (e.g., SPSS, R, SAS, Stata). It also offers an additional electronic book interface which allows the embedding of statistical operations (including those performed in other software) into an electronic text book environment. Stat-JR has been developed through work packages in 3 ESRC grants (2 NCRM nodes and 1 node in the Digital Social Research programme).
There is however a very important area which was not previously covered by our funding and that is the use of statistical software generally (and the use of Stat-JR eBooks specifically) for teaching introductory level statistics and software skills. As our funding has been primarily from the ESRC, our emphasis has generally been on pushing the envelope on what complex statistical models we can estimate and make available to the research community. We have of course a history, through for example our LEMMA training materials, of creating teaching resources for multilevel modelling (accessed by well over 15,000 registered users, but almost exclusively by researchers). We now see an opportunity with the increase in emphasis in quantitative methods at undergraduate level of producing interactive eBook based materials and indeed a system to aid with production of bespoke quantitative methods teacher-specific material using the currently most popular statistics software package for this community, SPSS.
In this project we have improved on the interoperability between Stat-JR with SPSS. We have also extended and improved the ease-of-use of the eBook writing functionality contained within the TREE Stat-JR module. Using this new functionality we were then able to produce a series of Stat-JR templates that generate outputs suitable for use in teaching practicals.
Over the course of the project the following twelve topics were identified. For each of these a generic eBook, which the user can use with their own data, and an example using the PISA data was created. Outputs from the PISA PDFs are linked below:
- Descriptive statistics for categorical variables [Practical] [Quiz]
- Descriptive statistics for continuous variables [Practical] [Quiz]
- Tabulation [Practical] [Quiz]
- Tests for normality [Practical] [Quiz]
- Independent samples t-tests [Practical] [Quiz]
- Paired t-tests [Practical] [Quiz]
- Non-parametric tests for unpaired [Practical] [Quiz] and paired data [Practical] [Quiz]
- Chi-squared tests [Practical] [Quiz]
- Correlation [Practical] [Quiz]
- Linear regression [Practical] [Quiz]
- ANOVA [Practical] [Quiz]
- Multiple regression [Practical] [Quiz]
For details on how to create your own versions of these see the resources on the Stat-JR downloads page.
- Progress on the British Academy funded Statistical eBook grant. Talk to the British Academy's Higher Level Strategy Group for Quantitative Skills by Bill Browne, Chris Charlton and Liz Washbrook, September 2017
- The British Academy funded Statistical eBook grant. Talk to the Q-step co-ordinators group at the Nuffield Foundation by Bill Browne, Chris Charlton and Liz Washbrook, November 2017