The original question was posted to Allstat on Friday, October 15, 1999:
*********************************************************
Up until now, statistical support in GKT School of Medicine (formerly UMDS)
has been largely covered by a single grant. There has been no charge for the
individual projects. In future, this may have to change.
I would appreciate any advice on how to charge for statistical advice. How
do you approach it within your institution? One approach is to expect the
cost of statistical advice to be written into any grant proposals. Is this
workable? Is 3% of the total figure reasonable? More? Less? If not,
how can the rates be set? By size of project (number of subjects,
number of repeat measurements)? Should we be charging by the hour?
Any other ideas?
I received fifteen responses, all from statisticians with experience of UK
Higher Education & Research Institutes, eleven being explicitly medical.
These are summarised below. One further respondent from the University
of Birmingham dealt with a slightly different question: an advice service
to two nearly local Hospital Trusts. Statisticians from Dunedin School
of Medicine, New Zealand and the University of Wisconsin, USA also
replied.
All fourteen respondents who mentioned it expected research proposals
to include statistical funding. At de Montfort University, the statistician
writes a mini-proposal. At Keele University and the Northern General
Hospital, Sheffield, each project funds one-quarter to one-third of the
costs of a full-time statistician, spread over the total project time. In
this way, three or four projects can fund one statistician. This allows
the statistician to contribute to teaching and administration in the
normal way.
Funding may be per hour, per month, or a fixed rate for all work on
a project, agreed in advance. The rates mentioned were £45 and
£50/hour (twice each) and £300/day. Other important points included:
· set charges according to the quality of the statistician needed,
· allow generously for overheads (PHLS allows 29%)
· have an "agent" who deals with all the non-scientific aspects - contracts,
pursuit of payments, deadlines, resolution of disputes
· Charge a higher commercial rate (£100/day) where appropriate.
Several respondents talked of the difficulty of getting medical researchers
to accept the cost of statistical support. Some told anecdotes of strange
behaviour:
· One medical research team took 2 weeks to produce a flawed analysis
that was rejected by journal reviewers. They were then unwilling to
pay for two days statistical work that lead to the paper being accepted.
· When statistical work took less time than expected, another team
expected the statisticians to make up the hours with envelope stuffing
and data entry.
Only one institute (Keele University) mentioned special terms for
students - £25 per hour instead of £50. Five offered first
consultations free, particularly where this formed part of the
process of drawing up a grant.
Imperial College School of Medicine gave the most detailed account.
Their Statistical Consultancy Service is described at
http://www.med.ic.ac.uk/df/dfsh/consult.htm.
They employ six statisticians plus two part-time support workers.
They receive a partial subsidy, covering accommodation and the
salaries of the two most senior statisticians.
ISCM has a very strong statistical base. The service is
part of the Department of Medical Statistics and Evaluation, headed by
Professor Simon Thompson and Professor Patrick Royston, with 22
other staff.
Paul T Seed MSc CStat ([log in to unmask])
Department of Public Health Sciences,
Guy's Kings and St. Thomas' School of Medicine,
King's College London,
5th Floor, Capital House 42 Weston Street, London SE1 3QD
tel (+44) (0) 207 848 6621
fax (+44) (0) 207 848 6620
web page (useful Stats handouts):
http://www-phm.umds.ac.uk/Statinfo/other.htm
Tuesdays only:
Public Health Medicine, 11th Floor, North Wing,
St Thomas' Hospital, London SE1 7EH
tel (44) (0)171 928 9292 x 1511
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|