Back to MEANS Homepage Participants Elizabeth Coates (Sheffield Hallam
University), David Chant (University of Essex), Neville Davies (Nottingham
Trent University), Michael Dewey (University
of Nottingham), Simon Dunkley (SPSS), Anne Hawkins (RSS Centre
for Statistical Education), Ann Hewkin (recent graduate now at
Knoll Pharmaceuticals), Peter
Holmes (MEANS Project), Alan Irving (Zeneca),
Michele Lundy (CCN), Margaret Rangecroft (Sheffield Hallam
University), Ed Redfern (University
of Leeds)
Notes
Peter Holmes and Anne Hawkins welcomed all those attending and
gave a brief background to the MEANS Project, outlining what
it hoped to do and spelling out the purpose of this and other
workshops. They both described different aspects of some of the
findings so far from correspondents, visits and from following up
job advertisements.
In a free-flowing session individual participants were
encouraged to describe their backgrounds and the issues involved
in improving the match between undergraduate statistics courses
and the needs in employment as they saw them. (Everyone was
encouraged to report on positive as well as negative experiences).
The following notes are based on the personal experiences and
perspectives of the workshop participants.
Michele Lundy (CCN), picking up on an earlier comment
about the use of practical work in undergraduate courses, said
that the sort of practicals included in university courses did
not necessarily match up with the practical problems encountered
in industry. Learning had to take place in work after people had
left university. In general the MEANS Project may have to
distinguish between what employers want and what the university
would give - maybe we had to identify and classify the different
ways that statistics is used since different employers would have
different needs. This might lead to different versions of courses
or more choice within courses.
She was looking to students having more training in
presentation skills and a real grounding that is appropriate to
later needs. Some of the problems arise from employers not saying
what they want. Students may need to realise that some things
they learn will only fully make sense later. The ability to ask
right questions only comes with practice, though this could begin
at university. She found many graduates at BSc level appeared to
have been spoon-fed. Those who had had periods of placement
during their courses seemed to be better at asking intelligent
questions; they were more mature in outlook.
Others commented that they found some undergraduate courses so
sanitized that graduates found real questions difficult. There
was support for the view that maturation came from placement
courses and a comment that no-one fully understands anything
until they have used it. Michele asked for more links between
users and academics - she would be willing to give a lecture on
how she uses statistics (together with an assignment) and would
welcome contact with lecturers to be kept up to date with current
thinking and statistical packages.
Alan Irving (Zeneca) had previously worked for Rolls
Royce and done some university lecturing. He had always had to
solve practical problems in his work and often had to work one-to-one
with a customer. He had tried to use this technique in
interviewing graduates for jobs but, on the whole, it was not
successful as the graduates could not put themselves into the
position of either the consultant or the customer. This may have
been as much a problem with personality as with education. The
problems the graduates had met as students had not prepared them
for the real problems of work. Since many statisticians have to
work in teams it would be useful if university courses could
introduce students to groupwork, group interaction and
brainstorming. In his experience the statistician can often see a
particular problem in a more global context.
Simon Dunkley (SPSS) had a background in psychology
rather than statistics. Many of his fellow students did not get
past the first lecture in their statistics course which was on
levels of measurement. His undergraduate course had had a
cookbook approach and did not convince him of why statistics
might be useful to him. It was only when he did a postgraduate
course in statistics and IT that things began to click. He is now
part of the training department at SPSS where he is running
courses for non-statisticians as well as statisticians. He finds
he is trying to help his customers develop statistical thinking.
Are statistics degrees just creating more mathematicians? If so,
this is not a good starting point for the applied statistician.
Statistics students need to know the why, what and how of
statistics in the context of solving problems. They need to know
generally what they should be looking at. For example - look for
shape, bias and precision, develop a feel for distributions, look
for numbers to mean something in a pattern. How we teach
statistics will be different for the mathematician and the non-specialist.
The first steps are to understand why we are doing statistics and
to put everything in an over all framework. In training courses
he is careful not to focus too much on evaluation but to
challenge people to think, interpret and check that their answer
really does answer the question posed.
University courses need to have the right balance between
content and understanding and to make best use of the time
available.
Michael Dewey (University of Nottingham), is involved
with postgraduate and undergraduate statistics service courses at
the University of Nottingham. In these courses there is little on
problem solving. He would like the students to develop
interviewing and consultancy skills - particularly in questioning
- and learn general methods of thinking through a problem.
David Chant (University of Essex) has recently been
working in Australia and New Zealand. Looking at who employs
statisticians he has found that there are some companies who
expect to train and others who expect the statisticians to train
themselves and ironically this did generally not accord with the
needs of recruits. In the context of social sciences, many
statistics courses are more mathematical than the students expect.
Perhaps they need a skill and they can be re-trained if needed.
Maybe it would be more sensible for such students to receive
their initial statistics courses outside the mathematics
department. The statistics needed in sociology and psychology is
not overly mathematical.
Mathematical skills are not particularly relevant in many jobs
taken by social science graduates. Current advertisements seem to
refer to the use of databases, report writing, communication and
computing skills using packages such as SPSS and SAS.
Universities should deliver what makes a student employable. MEANS
should get on board people who teach statistics from outside
mathematics departments. Good pedagogy should lead to students
finding it easier to get good jobs.
Neville Davies (Nottingham Trent University) described
how first year service courses were often constrained by other
departments' ideas of what should be in second year courses. He
has a service course for economists which has 190 students. He
uses technology in teaching and has some students make quick
evaluations at the end of each lecture (what went well, what didn't
you understand, what needs improving). There is much variability
in the feedback comments. He uses the induction weeks to collect
data about the students which they then use in lectures.
Ann Hewkin (Knoll Pharmaceuticals) referred to her
recent undergraduate experience. Her first degree had been in
mathematics which was all based on pure theory with no numbers.
She felt that the lecturers were in a rut, with little enthusiasm.
Examinations seemed to be mere rote learning. Statistics formed
about one sixth of her mathematics degree. She then went on to do
an MSc in Medical Statistics, for which the basic mathematical
requirements were A-level mathematics. This course covered little
theory, was mostly relevant applied examples but did not try to
answer the deep down 'why' questions. This course was enjoyable
and was the basis of her getting her present job as a clinical
statistician. The job requirements as advertised were; a degree
with reasonable statistics content or an appropriate Masters
degree. She felt that she could have applied after her first
degree but would not have been able to answer questions at
interview. She needs to use SAS in her job and did have a short
course on it although most was learned from colleagues. After she
had completed her MSc she went back to look at her undergraduate
mathematics lectures and found them useful - but the two had not
merged together before this.
Margaret Rangecroft (Sheffield Hallam University) does
most of her statistics teaching in non-statistics departments.
There is more flexibility in these courses than those for
specialist statisticians. Within the statistics degree courses
she finds that students are reluctant to 'do' statistics, they
prefer theoretical approaches. They do not want to think about
problems. They prefer spoon-fed theorems in a structured
theoretical course. An attempt to include real situations in
examination questions was resisted by the external examiner. In
the third year the students go out on placement and this makes a
lot of difference to their approach and maturity when they return
for the final year. In the second year, these students also do
group work as a project for external clients. Throughout the
course, students have half a day each week on practicals in
different areas. Mock interviews are arranged for the students -
but they tend to get uptight about these.
Elizabeth Coates (Sheffield Hallam University) said
that all students on the Applied Statistics degree had to take a
unit in the first year which is designed to enhance their
communication skills and their knowledge of business, set as far
as possible in the context of the sort of environment they are
likely to find during their placement. The statistics units also
include oral presentations. The emphasis in the statistics units
is on asking the right questions, probing to clarify the real
nature of the problem posed, searching for an appropriate
technique, ensuring that the chosen technique is really valid,
interpreting and communicating the results.
Ed Redfern (University of Leeds) is part of the
directorate of a consultancy unit which gives insight into the
nature of skills needed by employers. They are trying to
incorporate more technology into the teaching - they use the
STEPS material. In the first year course in Pure and Applied
Statistics there is group discussion; students present material
to each other and mark each other. The problems can be fairly
open-ended e.g. how to discriminate between two types of item (the
STEPS example is how to tell the sex of a seagull). Some of the
staff feel that there is a need to have the students working in
groups and then the students share out the group's mark according
to the degree to which each person has contributed. There is
resistance to this from some other colleagues who argue that this
is not statistics and not mathematics. Ed sees that, through the
groupwork, students teach each other. There is a problem with
class size which has risen recently from about 60 to about 100.
All use Minitab and SPSS or SAS. All will have written a report
on what they have done. Practicals have to be tailored to the
time available.
All this side of the teaching takes a lot of staff time. Staff
have to assess individual reports, the group reports, the posters
they prepare and the content of them. This is not part of the
final year where there is a more diverse choice of topics and
students are more jealous of their individual grades.
Neville Davies (Nottingham Trent University) described
a second year course he had just introduced called Communicating
Statistics is which he looks at statistics in the media and
how people communicate statistics. He gets outside speakers to
take part in this course and he has put his notes on to his Web
page. With this course he has established an email mentor system.
The mentors agree to answer students' questions about their daily
statistical activities. This is intended to get the students
talking to professional statisticians.
Some final comments were made on the ability of students to communicate and the need to develop listening skills as well as to ask questions. Although not all job advertisements mention communication skills it is unlikely that they won't be needed.
The meeting split into two groups to consider what were the
influential factors on universities and how courses might respond
to the needs in employment.
The influential factors are: employers, students, academics,
examiners and, to some extent, recruiters. The universities are
trying to educate the students so that they are generally
employable. The employer will require specific skills which may
range from simple technical skills to advanced thinking for
Research and Development. Additionally it is useful for students
to have general skills such as the ability to learn, to ask
questions, to communicate, to work in a team and to be able to
evaluate critically a piece of work. Simple statistical things
such as EDA, reading tables, interpreting data should be at the
core of all courses and be reinforced throughout the course.
Universities do not produce the finished product, learning goes
on throughout life. It would be useful, therefore, for
statisticians employed outside education to be kept in touch with
academic statisticians both because people forget over time and
need their knowledge refreshing and because new ideas and
techniques are being developed.
If change is to take place, then there is a need to get more
people involved and wanting change. We have an internal marketing
problem within the universities. Cross fertilization between
courses may help. Outside lecturers on existing courses might
help to bring a broader perspective. There are many obstacles to
change: size of classes, modular systems, examinations, and
general extra stress on university lecturers.
P Holmes
November 1996