This miniproject aimed to develop and evaluate an extensible support resource to encourage independent learning of statistical topics.
The expected outcomes of the project included:
1. the production and web-publication of a pilot R-book interactive resource covering some core statistical concepts, suitable for use in guided learning; and,
2. a student evaluation of the way this resource can be used in learning.
A wide range of statistical topics were identified by survey as being of interest to transition-stage mathematics students. Nevertheless, a number of these are already well served by existing resources (the STEPs project for example dealt with experimental design and the rpanel work deals very well with Ordinary Least Squares regression). Consequently, further workbooks were added for:
1. practical interpretation;
2. bootstrapping; and,
3. confidence intervals.
Feedback was obtained using the web-based survey tool at surveymonkey.com. Due to the timing of this project, most of the initial feedback was from non-target students (for example, undergraduate medical school students, postgraduates on development courses). The feedback was positive regarding the usability of the resources. Nevertheless, the project results will be ‘polished’ by gathering feedback on these R books from transitional mathematics students.
Assessment resources were developed using Acrotex and R. The idea is that R can process an Acrotex template file and generate customised data sets, along with a matching worksheet for each student in a class. Students then get a range of guided activities, with customised data, and Acrotex offers the potential to mark their work.
In R, Sweave() provides an implementation of Donald Knuth’s noweb idea. It is possible to combine conventional LaTeX with R code.
For example, one can add:
<<BitOfCode, fig = TRUE, echo = FALSE, results = tex>>= xtable(xtabs(~x+y))
@
within some piece of LaTeX.
The full text file can be run in R though Sweave() whereby the R code (xtable(xtabs(~x+y))) is executed and the results placed inline within the text file replacing the inclusion. The resulting text file can then be processed using LaTeX.
In order to tighten up the assessment potential of this project, David Story’s Acrotex bundle was used with the Sweave facility. This means that it is possible to create a number of essentially similar assessment worksheets, but they will have subtly different data contained within them. This has clear potential in terms of plagiarism, but is in itself instructive when dealing with large sample statistical theory such as confidence intervals and the central limit theorem. Essentially, an end user can submit a class list to an Rbook, and receive back a data file and associated PDF worksheet for each member of the class. Each worksheet will be slightly different.