Department of Mathematics

Statistics advice centre

For many years the Statistics Group, operating within the Department of Mathematics, has offered a statistical analysis and advice service to commercial organisations, as well as to internal researchers and postgraduate students at the University of Surrey.

Contact us

To make an appointment with the Statistics Advice Centre, you may:

The Statistics Advice Centre can be found on Level 4 of Block AA, next to Senate House.

Research grant application advice, sample size calculations

The Statistical Advice Centre is able to offer guidance on the calculation of required sample sizes for proposed research. This is an essential feature of any grant application and is examined rigourously by ethics committees such as the Animal Welfare Ethics Review Board (AWERB) at the University of Surrey.

The National Institute for Health Research (NIHR) funds the Research Design Service (RDS) to provide research design and methodological support to health and social care researchers across England to prepare applications in applied health research, for submission to the NIHR and other national peer reviewed funding programmes. Advice is confidential and free of charge. NIHR Research Design Service South East (RDS SE) covers the counties of Kent, Surrey and Sussex; the Surrey site is based in FHMS at the University of Surrey.

Statistical advice from the RDS SE is provided as part of a package of wider support. For further information, and to register for support visit http://www.rds-se.nihr.ac.uk/

For informal enquiries, please contact Kay Stephenson k.stephenson@surrey.ac.uk

Please note that the RDS cannot provide support to students (undergraduates or postgraduates). Students in need of research support and advice should liaise with their supervisors.

Should a proposal (typically a non-clinical one) not be eligible for RDS SE input, more limited free advice (typically in relation to quantitative statistical issues such as required sample sizes) is available directly from the Statistics Advice Centre to University of Surrey staff and postgraduate students, dependent upon the Faculty/Department of the Lead Investigator.

Please email stats-advice@surrey.ac.uk for further details.

Academic & external research

For full details of the academic research work of our Statistics personnel, please see the Statistics Research Group's web page.

In addition, collaborative research takes place with personnel from many other departments. A recent example involves nutrition personnel in the School of molecular and Biological Sciences, to whom we provided expert assistance on stepwise logistic regression. This enabled the identification, from a substantial data set on the feeding patterns of a large number of premature babies, of a model to predict the incidence of eczema within 12 months from birth. The appearance in this model of controllable items such as number of different solid foods led to the derivation of feeding recommendations for improved health in premature babies. Furthermore, publication was achieved in a high ranking academic journal.

Rewarding collaborations with external parties also take place frequently. For instance Cephalon, a local biopharmaceutical company, approached us for statistical assistance in analysing the results of an extensive clinical trial. From their point of view, access to our Statistics personnel provided them with expert statistical analysis without the commitment of the long term employment of a professional medical statistician. From our point of view, exposure to current practices in drug safety trials expanded our knowledge of statistical procedures required in the commercial world.

Unfortunately owing to the recent high level of demand for statistics advice within the University, we are unable to provide help to external individuals and companies at the present time.

Research project services

The service is able to perform a full range of tasks associated with all aspects of a research project. These include:

  • Quotes for statistical analysis for inclusion in project proposals
  • Statistical advice on required sample sizes
  • Assessment of study protocol and of the possible need for a pilot study
  • Statistical reviews of previous studies and of relevant literature
  • Advice on design of questionnaires/clinical report forms
  • Data collation
  • Construction of data coding schemes
  • Data entry : advice on the use of an appropriate spreadsheet/database
  • Data validation (checking data for errors)
  • Statistical analyses, including graphical output if required
  • Provision of bespoke statistical software, including documentation

Whilst the amount of help we can give on some items may be limited (for instance the important issue of sample size is often determined purely by logistical considerations), we are always available for short sessions to offer advice on any of the items above. In particular, a brief review of your questionnaire design may lead to enhancements which will enable a much more incisive statistical analysis after the data have been collected.

Finally, please be aware that all Statistics Advice Centre staff have other work commitments, ranging from teaching related duties to involvement in major research projects. Consequently, although an initial response to any statistical query should arrive within a few working days, it may occasionally be several weeks before the staff member who is most suited to the query is available for a face to face meeting. Consequently, planning ahead for the statistical element of any project is essential.

Statistical computing

Following data entry and validation, statistical analyses are performed on platforms ranging from a Hewlett Packard 9000/870 mainframe running HP-UX (Unix) to Macintoshes and PCs running Windows. The Statistics Advice Centre has a range of statistics software packages at its disposal, ranging from simple epidemiological and graphical products, such as Excel, Epi-Info, CricketGraph and StatView, through to sophisticated statistical packages, such as SAS, SPSS and Minitab. SAS is the product of choice for most projects, with its broad catalogue of statistical procedures and excellent graphical capabilities. Together with its Applications Facility for constructing menu-driven screens, which can be tailored to the requirements of the client, these provide the most suitable and user-friendly working environment. The Statistics Advice Centre also has experience with statistical products such as GLIM, SPLUS, GENSTAT, XLISP-STAT and FORTRAN with NAG routines, and can thus usually accommodate a client's specific software requirements.

Organisations that have made use of our statistical computing service have included pharmaceutical companies, medical research institutes and government agencies.

Clinical trials

The clinical trials management service performs a full range of tasks associated with all aspects of medical research. These include:

  • Quotations for statistical analysis for inclusion in project proposals
  • Statistical advice on required sample sizes
  • Assessment of study protocol and of the possible need for a pilot study
  • Statistical reviews of previous studies and of relevant literature
  • Advice on design of questionnaires/clinical report forms
  • Data collation
  • Construction of data coding schemes
  • Data entry, or advice on the use of an appropriate spreadsheet or database should the client prefer to perform data entry in-house
  • Accessing large computerised databases for data retrieval
  • Data validation

Following data entry and validation, statistical analyses are performed on platforms ranging from a Hewlett Packard 9000/870 mainframe running HP-UX (Unix) to Macintoshes and PCUs running Windows. The Statistics Advice Centre has a range of statistics software packages at its disposal, ranging from simple epidemiological and graphical products, such as Excel, Epi-Info, CricketGraph and StatView, through to sophisticated statistical packages, such as SAS, SPSS, GLIM and Minitab. SAS is the product of choice for most projects, with its broad catalogue of statistical procedures and excellent graphical capabilities. These features, together with its Applications Facility for constructing menu-driven screens (which can be tailored to the requirements of the client), provide a suitable and user-friendly working environment.

Extra information

Consultancy charges for external clients

As of 1st December 2013, all statistical advice to external medical researchers is charged for at 60 pounds sterling per hour, or 450 pounds sterling per day, + VAT . Note that VAT is not chargeable if the work is being paid for from within the University, however VAT will be charged for all external consultancy.  

External companies are charged 600 pounds sterling per day + VAT (or 80 pounds sterling per hour + VAT), with substantial discounts (up to 20%) for projects requiring more than 5 days of work.

Additional charges are made for travelling time and expenses, should meetings be necessary outside of the University. At the first consultation the total number of hours will be estimated, according to the extent of the required data handling, statistical analyses and report format. An invoice is dispatched upon completion, with payment requested within 4 weeks.

Note however that assistance to external companies is subject to staff availability.

Suggested checklist for attendances at statistical advice meetings

  1. MOST IMPORTANT OF ALL : a copy of the original questionnaire/result sheet
  2. A list of clearly defined objectives for the project
  3. Copies of similar studies for reference (if applicable)
  4. A copy of the latest version of the data (if applicable), saved as
    1. an SPSS ".por" export file (NOT an SPSS save file), or
    2. an EXCEL file (version 4 or higher)

Literature

There are many good books on statistics, and just as many bad ones. If you are looking for a well written text, below are some suggestions.

Statistics without tears
by Derek Rowntree (Penguin) £9.99   ISBN9780140136326
A good introduction to statistics for those with a minimal background in mathematics

Statistics explained
by Perry Hinton (Routledge) £26.99   ISBN9780415332859
A very good introduction to all of the common statistical techniques

SPSS Survival Manual
by Julie Pallant (Open University Press) £36.99   ISBN9780335261543
Highly popular guide to using SPSS software package for simple statistical tasks

And for more specialised branches of statistics:

Practical Statistics for Medical Research
by Douglas Altman (Bookpoint) £38.99   ISBN9781584880394
Covers all aspects of medical research, with special attention to the requirements for ensuring that papers are publishable

An Introduction to Medical Statistics, 4th edition
by J Martin Bland (Oxford University Press) £29.99   ISBN9780199589920
Excellent guide to medical statistics

Applied Regression Analysis and Other Multivariable Methods
by Kleinbaum, Kupper and Muller (Cengage) £79.99   ISBN9780495384984
Excellent text on all aspects of regression analysis, with SAS code

Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences
by Jacob Cohen, Patricia Cohen, Stephen West and Leona Aiken (Lawrence Erlbaum Associates) £75.00   ISBN9780805822236
Excellent text on all aspects of regression analysis, with SAS and SPSS code

Finally, for an excellent general introduction to all aspects of carrying out a research project, including basic statistics:

Research Methods in Psychology, 4th edition
by Glynis Breakwell, Jonathan A. Smith and Daniel Wright (Sage) £35.99    ISBN9780857022646

 

(Prices quoted January 2017)

Articles for publication

When collaborating on research papers, co-authorship is not normally necessary, although an acknowledgement for statistical assistance is always appreciated. Co-authorship might be suggested, however, where statistical support is complex and/or substantial.

Recent examples of medical projects

In addition to clinical trials, the Statistics Advice Centre undertakes many other types of project associated with medical statistics. Here are some recent examples:

  1. A clinical trial to compare the efficacy of 2 wound dressings: The British subsidiary of a large American pharmaceutical corporation required a comprehensive statistical report on a clinical trial comparing the effectiveness of 2 different types of wound dressing in the treatment of leg ulcers. Data entry, data validation, and statistical analyses, including graphical displays, were performed.
  2. Construction of a push button statistical reporting system: In order to deal with frequent requests from internal medical research staff at the University of Surrey, a push-button statistical reporting system was developed to handle efficiently the most common types of analyses required.
  3. Simultaneous interrogation of five downloaded data sets: A major research project was undertaken to assess the workload at clinics for sexually transmitted diseases in the light of the AIDS epidemic. A large and complex database was downloaded from a major London clinic's patient administration system. After intensive statistical analysis the most accurate marker of the clinic's increasing workload was identified. The results were published in an appropriate medical journal.

Recent examples of computing projects

  1. A clinical trial to compare the efficacy of 2 wound dressings: The British subsidiary of a large American pharmaceutical corporation required a comprehensive statistical report on a clinical trial comparing the effectiveness of 2 different types of wound dressing in the treatment of leg ulcers. Data entry, data validation, and statistical analyses, including graphical displays, were performed. [Software/hardware configuration : SAS on the mainframe, Excel on the PC]
  2. Construction of a push button statistical reporting system was developed to handle efficiently the most common types of analyses required. [Software/hardware configuration : SAS/Windows on the PC]
  3. Simultaneous interrogation of five downloaded data sets: A major research project was undertaken to assess the workload at clinics for sexually transmitted diseases in the light of the AIDS epidemic. A large and complex database was downloaded from a major London clinic's patient administration system. After intensive statistical analysis the most accurate marker of the clinic's increasing workload was identified. The results were published in an appropriate medical journal. [Software/hardware configuration : SPSS on the mainframe, Cricketgraph on the Macintosh.]