Abstract:

We report on an open access statistics resource created for the UK national statstutor site. This site is the first port of call for students and lecturers looking for post-16 statistics support materials. The resource described here comprises a suite of e-Assessments together with supporting materials that relate to the statistical activities involved in choosing and carrying out an appropriate one sample test for location (mean or median) on a randomly generated data set. Five e-Assessment modules are available and these may be accessed independently or can be taken sequentially mimicking the flow of a full statistical analysis using the SPSS software package. The goal is to identify and perform the appropriate location test for a particular data set; t-test, Wilcoxon signed-rank test or Binomial (Sign) test.

The e-Assessment modules are as follows:
1: Exploratory Data Analysis
2: Shapiro-Wilk Normality test and Skewness & Kurtosis
3: One-sample t-test
4: Wilcoxon signed-rank test
5: Sign (Binomial) test

The e-Assessments are run using DEWIS which is a fully algorithmic open-source e-Assessment system designed and developed at the University of the West of England, Bristol (UWE). It is a completely stand-alone web based system used for both summative and formative assessments. DEWIS uses embedded R code in order to create bespoke random data sets with desired statistical properties and to generate answers that match SPSS screen output. On accessing each e-Assessment module you are able to download a new statistical data set or, by using a unique customer reference number, return to a particular data set that you have previously used. Each module requires you to download this data, run it on the SPSS statistical package, get output and answer a few questions demonstrating understanding of the results. Each data set is generated randomly and thus you can have multiple attempts at the same scenario with a variety of data sets that will cover the various experiences that you will encounter when analysing data; i.e. data that is appropriate for any one of the three tests and that may or may not have a significant effect.

This resource is relevant to a wide range of disciplines, where it is necessary to master a methodical and defendable approach to carrying in depth and appropriate statistical analysis. A variety of application contexts are offered, for example from the fields of business, environmental science and engineering. Data sets are generated that vary in sample size, normality and in their degree of skewness. It is recommended that the accompanying notes are downloaded before using the resource for the first time, which describes the rule of thumb guidelines that enable choices of the appropriate location test based upon sample size, outlier presence, skewness values and normality test results.

Feedback is given for inappropriate test choices as well as other incorrect analysis. Videos and instruction pamphlets are accessible as links from each e-Assessment, which give clear instructions as to how to carry out the analyses and interpret results using SPSS. These videos and pdf instruction pamphlets are also available as stand-alone documents. Repeated use of the e-Assessment modules facilitates learning of how to identify and employ the correct test on a variety of data sets.

Keywords:

E-Assessment, statistics, SPSS, multimedia, higher education.