network for excellence in
    mathematics and statistics support

Discussion points from MSOR statistics session

Introduction from Maths and Stats Help (MASH) Sheffield

Sheffield University currently has 8 statistics tutors, including myself.  The majority are PhD students from either the statistics department or the School of Health and Related Research.  Stats sessions are run every morning in term time and twice a week during holidays.  We find that the majority of students require help with project statistics rather than course material.

Key discussion points:

  1. 1.     Supporting rather than consultancy

One of the main problems with statistics support is that students come with their own data and often require help choosing an appropriate test, carrying out the test and interpreting the output.  This is a particular problem with dissertation students.  In response to this, we decided to look for example data for our tutors to use and were hoping to produce example sheets from the data for students to use.  See ‘Datasets resource proposal’ (1) below for more information.  One possible problem with this is that the students’ data may not meet the assumptions but when our data does.

  1. 2.     Tutor background

Students from all backgrounds come in for help and each discipline has its’ own commonly used statistics, some of which we know little if anything about.  This can lead to some unhappy customers if none of us can help.  One university uses PhD students within each discipline rather than those from a statistics background to deal with this problem.  One concern of this is that the tutors may not have been taught statistics correctly.  These students are given some training before becoming advisors though.  Another problem is knowing how to carry out analysis in the chosen statistics package and interpret the output.  None of our tutors have been taught SPSS formally.  We have started giving basic training when tutors start and are lucky to have more than one tutor for each session so between us we can usually manage.  Could more training be given to new tutors and if so, what would be most helpful?

  1. 3.     Supporting centres with little or no statistics support

Leading on from the previous section, we are aware that there are other universities with very little support available.  Group discussion came up with the following thoughts:


  1. 4.     Online material and student feedback

The production of online material is time consuming but how do we know if students want or need it?  We attempted to monitor the use of our website for a few months to see what the most popular topics were.  Currently, we use the HELM material and worksheets from Coventry and Loughborough (available on statstutor) as well as some of our own material and numerous links to other websites.   Although there were problems with the methods of monitoring website use, the top downloads appeared to be on choosing the right statistical methods, descriptive statistics, ANOVA and the normal distribution.  What we don’t know is whether the student found what they were looking for and if it was useful.  I have asked for feedback on worksheets given to students but heard nothing so far!  If anyone else has student feedback regarding online material and/or worksheets, we would like to hear about it.  Also, should we produce our own material or use other websites?  Is it preferable to combine peer reviewed resources as a statistics support community perhaps via Statstutor?  What are the advantages/ disadvantages of this?  At the moment, my thought is that concentrating on producing more material on choosing the right test (e.g. in flow chart form) would be helpful for students and tutors.  We have started doing this at Sheffield and believe that at least a couple of other people are working on something similar.  It’s not as easy as it sounds and combining the knowledge of several people is useful.

  1. 5.     Encouraging better planning of studies

It is most common for students after they have collected their data which can cause problems.  How can we encourage better planning of studies?  One university is holding sessions in September for new PhD students.  We hold workshops on request for groups of students and have considered holding sessions for postgraduate students but are concerned about overlap with what is already available and stepping on other people’s toes.  Also, general SPSS sessions for example, may help reduce repetitiveness in one-to-one sessions.




(1)   Datasets resource: This is a relatively new project involving Sheffield, Leeds, York and Warwick universities.  The aim is to create an online resource containing a few datasets which can be used to demonstrate numerous statistical tests.  There are a lot of datasets available on the web but they tend to contain just the two variables needed for that one statistical test.  Part of the project involves setting up web pages making it easy for a tutor to find an appropriate dataset to use and even suggested hypotheses to test.  It is hoped that eventually this resource will be contained within the new statstutor website and will contain a selection of discipline specific datasets.

(2)   Statstutor: This is a new statistics resource set up by Alun Owen of Loughborough University.  It is the sister website to Mathscentre and is laid out in the same way.  Currently worksheets used by Loughborough and Coventry universities are available and other peer-reviewed material but more is expected to be added in time.