Good practice for formative
assessment and feedback in statistics
MSc(R) thesis, University of Glasgow.
Full text available as:
Feedback to learners about their work is an important part of the teaching and
learning process for any subject. Feedback should ensure students are clear on
where they went wrong and what they can improve in the future. Without useful
feedback students continue to make similar mistakes. However mathematical
subjects such as Statistics appear to place less emphasis on feedback compared
to other subjects. Statistics has made steady pedagogic progress and now uses
a variety of assessment methods, but producing effective feedback for these
methods has not made the same progress. This thesis investigates the feedback
currently given to some Statistics classes for students who are studying Statistics
as part of a degree in another subject, proposes a set of useable guidelines for
producing effective feedback and reports on the creation and piloting of a
multiple choice, computer-aided assessment system that provides immediate
feedback to learners in Statistics courses.
The first chapter of the thesis discusses the background of the subject. Key
features include the Quality Assurance Agency’s code of practice, which
institutions should be following with regards to assessment, the National Student
Survey, in which results for assessment and feedback are generally not
favourable, and the various models for Statistics assessment suggested by Gal
and Garfield in their book The Assessment Challenge in Statistics Education. An
interesting thing about this book is that, though the whole book focuses on
assessment, there is little mention of how to give feedback for any of the models.
Chapter two reviews the literature on feedback. This reveals that feedback can
improve or impair performance depending on various factors. A summary is
given of the most repeated guiding principles for constructing feedback. How
students use feedback, including guidelines for receiving feedback, is also
discussed. The final part of the chapter looks at the advice given for constructing
multiple choice tests and the lack of guidance for feedback relating to multiple
Chapter three describes student questionnaires that were implemented in
Statistics courses at the University of Glasgow to survey student attitudes to the
feedback they received. A questionnaire and follow up questionnaire based on
the guiding principles was piloted with a small group of second year Statistics
students. Before issuing the follow up questionnaire, the way feedback was
produced was changed in line with the guiding principles. When the
questionnaires were compared, students were more satisfied with how quickly
feedback was returned, the amount of feedback, the detail and the overall
usefulness of the feedback after the intervention. The questionnaire was then
adjusted to fit with a larger first year class. This included adding the Rosenberg
self esteem scale to measure students self esteem. These results showed that
the detail of the feedback given needs to be improved more than the amount.
The most common reasons given for why the feedback was not detailed enough
were that there was no suggestion for improvement, it was unclear where the
mark was lost and the feedback was too vague. There may also be a relationship
between students self esteem and the attention they pay to feedback. It appears
those with a lower self esteem pay less attention to feedback. At the end of
chapter three a briefing document is presented that can be used to help train
markers. This is a summary of the guiding principles and includes good and bad
examples of feedback.
Chapter four discusses the construction of a multiple choice testing system and
the creation of specific tests for use in a level one Statistics course. First the
chapter describes the piloting of another computer-aided assessment system
called Model Choice. The results of this were very positive, with all students
agreeing the system was easy to use and appreciating the immediate feedback.
Next a similar system was created for use with the Statistics class for
Psychologists and Social Scientists. Multiple choice questions were constructed
for four of this course’s labs, on sampling and interval estimation, multiple
regression, experimental design and categorical data. For each question, three
incorrect options and a correct option were produced. Feedback was also written
for each option explaining why the chosen answer was either correct or incorrect.
Students getting the answer wrong first time were then given a second attempt.
The literature on constructing multiple choice assessments was consulted during
Chapter five focuses on piloting the computer-aided assessment system. The
system was initially trialled with postgraduates and staff. The program received
an excellent response and a group discussion revealed plenty of constructive
ideas to improve the system. The program was then trialled with new third year
The final chapter summarises and discusses the results obtained to date and
makes suggestions for further work.
Actions (login required)