Back to MEANS Homepage I spoke mainly to people in the statistics department,
responsible for everyday activities, though I did meet the two
people responsible for Research and Development. MBI is one of
the largest market research organisations (worldwide not just UK).
Most of the people I met had come through Sheffield Hallam
University's Applied Statistics degree course which has a one-year's
placement. This one year out was influential in these people
being offered employment.
Chris
Recently taken over as manager of this section coming from AC
Nielson. There he had been working on the retail tracking of
store sales and other aspects of Market Research. He had come
through Sheffield Hallam University but did not recall any
marketing examples in that course. He recalled the course as
working from theory to practice, they did not do much on sampling
in a market research context; it was more linked to computing.
Currently he is working on building models to estimate store
turnovers. He left university seven years ago and had to teach
himself programming and the use of packages which he did using
SAS manuals. He was given support at Nielson to get to know SAS -
people were coming in with no SAS background and needed to
understand enough theory to be able to build models. He suggested
the universities needed to integrate genuine practicals into
their courses - perhaps they could get people from business in
for one day.
His work at MBI is more statistical than at Nielson. Nielson
tended to operate at store level data; at MBI they work with
respondent level data. He is concerned with how data are
collected, different types of data, the effect of images and
statements in advertising and general marketing data.
He has to put numbers to perceptions. [Is this sort of thing
done in statistics for business courses - need to check]. He
found his placement year prepared him well for work; it made him
more employable. He had work experience and a better
understanding of a business. His placement was at Nielsons where
he was given his first job.
To some extent his department at MBI have to teach statistics
to the rest of the company. This is at the level of elementary
statistics up to significance tests. For communicating with the
company he needed a broad education, especially when talking to
those with no statistical background.
Technically he uses discriminant analysis, CHAID analysis
based on different types of data. At the other end -
questionnaire design - a good course in this would have been
useful.
He continues to have to learn and is not sure where he learned
to learn. Some came through experience in trying to found out how
X is used.
Within the company they run occasional courses for 10/12
people to cover common questions that arise. They will also meet
1 to 1, 1 to 2 on request to help give individual insight into
more specialist analysis. There is an introductory course for new
employees then a specialised course on sales modelling for new
graduates. Here they look at the effects of advertising, price,
distribution on sales through a work example. This takes one hour.
They do keep these examples on file.
Dave (?) graduated 1989 from Sheffield Hallam
University on their Applied Statistics course. He is very much
the mathematician and programmer in the group. He likes working
through problems on his own and likes the challenge of a
difficult theoretical problem. He does not get as much involved
with communicating to non-statisticians an the others do, nor
does he meet clients. He worked for Grassland Research where he
was programming in Fortran on their Vax and using Genstat. Work
was linked to bivariate normal distributions and he was helped by
his background computer experience. Now he is using Visual Basic,
correspondence analysis, programs developed using Genstat code.
He taught himself CHAID from the software.
As an undergraduate he learned a lot of useful tools e.g.
eigen values for correspondence analysis. He has found the theory
useful in a wide variety of contexts. He remembers one compulsory
management course in his first year with a limited choice in the
final year. His final year project was analysing weather data (using
local data from Weston Park in Sheffield) and comparing forecasts
with actual weather. He used tests of independence (comparing
Leeds weather with Sheffield weather). He now needs cluster
analysis which was part of his multivariate course.
Correspondence analysis is important at MBI; it was only a small
part of his undergraduate course.
Sandrine is on a placement course here before going on
to Sheffield Hallam University. She was educated in France up to
the first part of her degree which she is finishing at Sheffield.
She did a one hour a week stats option in her baccalaureate. At
university she studied some statistics and probability options
and is now doing some computer science and statistics. This
includes modelling, sampling methods, SAS, macros, clinical
trials, data analysis (she did some correspondence analysis in
France). Her modelling course was too theoretical and not
sufficiently practical. In France she did one day a week work (by
mail) for a wine company on data about contamination by metals
and used Statlab for the analysis. All students did something
like this. Others worked for banks, gas companies, social
security, computer science companies, Post Office, public
companies. (Dick Gadsden or Kerstin of SHU will have more
detailed information).
Catherine Technically she uses correspondence analysis,
scatter plots, multiple regression (for sales and sales
prediction). Trade-offs and conduit analysis were not done before
she came to BMI - she learned from Paul Dyson of the R&D
department and also used manuals to teach herself. Her
undergraduate multivariate course was good grounding - she still
refers to her lecture notes - and the factor analysis was also
useful. She spent her year out at Smith, Kline Beecham analysing
clinical trial data. She had to write a report after each study
and this helped in getting a job. She had to learn to write in
language that people understand; she learned many things by
observing others. In her undergraduate course she did have mini-problems
where she had to report back to the group. She was not taught how
to see a problem as one capable of statistical analysis.
In retrospect she thought that she did not need the recap on A
level mathematics that was part of her undergraduate course and
thought that the computing component could be improved (she now
uses SPSS in her work). Good parts of the course included the
practical sessions and the report writing. For assessment she
would like to see a 25/75 practical/examination breakdown.
Peter Holmes
November 1996