UNIVERSITY OF HERTFORDSHIRE STATISTICS SEMINAR SERIES
Part of the Business School Research Seminar Programme
www.herts.ac.uk/business/staff_public/nhspencer_public/seminars
Supported by Department of Management Systems,
Business School, University of Hertfordshire.
Wednesday 12th January 2005, 2pm, room tba, de Havilland campus
Paul Marchant (Leeds Metropolitan University)
Title: "Meta analysis may be useful, but looking at the data always is"
Abstract:
"A major systematic review on a crime reduction remedy suggests that claims
for its effectiveness are justified. The review at first sight appears to be
an appropriate statistical synthesis of all relevant studies across the
world. However on examination, the statistical claims and methods are
unfounded. The talk will discuss the quite apparent problems in a simple
way. Essentially standard methods of statistics rely on assumptions, such as
that of independent random sampling. Things go horribly wrong if the
assumptions are not met. The review in question is potentially the tip of an
iceberg.
The talk will address general matters to be considered in the design of
evaluations and systematic reviews, so that any conclusions made are
'properly scientific' and therefore sound."
For further information regarding location, parking, etc.,
please contact Neil Spencer (details below).
**********************************************************************
Dr Neil H. Spencer
Principal Lecturer in Statistics
Director of Statistical Services and Consultancy Unit
Business School, University of Hertfordshire,
de Havilland Campus, Hatfield,
Hertfordshire, AL10 9AB, U.K.
Statistical Services and Consultancy Unit Contact Details:
Telephone: +44 (0) 1707 285574; Fax: +44 (0) 1707 285410
E-mail: [log in to unmask]
WWW: http://www.herts.ac.uk/business/centres/sscu
Personal Contact Details:
Telephone: +44 (0) 1707 285529; Fax: +44 (0) 1707 285410
E-mail: [log in to unmask]
WWW: http://www.herts.ac.uk/business/staff_public/nhspencer_public
**********************************************************************
|