Wednesday 7th November 2012

Statistics of academic performance

Darren Greenwood, (Division of Biostatistics, LIGHT, University of Leeds) [Presentation Slides]

Reliability of journal impact factor rankings

Journal impact factors and their ranks are widely used by journals, university researchers of all disciplines, and will be used either formally or informally in the imminent Research Excellence Framework (REF). However, these estimates and their ranks are often based on quite small numbers, and their precision is never considered. In this talk I apply methods to assess this that are more commonly used in education and in medical research. Bayesian Markov chain Monte Carlo methods are used to estimate the uncertainty associated with these journal performance indicators and to compare journals. Alternative methods will also be mentioned. I discuss to what extent decisions placed on journal impact factors are potentially misleading where the uncertainty associated with the measure is ignored and how this could be improved in future.

Brian Cattle, (Division of Biostatistics, LIGHT, University of Leeds) [Presentation Slides]

Maximising Revenue from the Research Excellence Framework (REF)

The REF is the new system for assessing the quality of research in UK higher education institutions. It replaces the Research Assessments Exercise and will be conducted in 2014. With a desire to maximise the potential financial returns within the specific submission rules for research output, we have used Statistical methods to devise software to maximise the financial return, or the grade point (star rating) of the overall submission within the unit of assessment. Because the final submission of many authors is not yet fixed, we also required methods to examine various contingencies and the effects that these might have on the REF income. Furthermore, because each paper submitted may have more than one author in the same unit of assessment, it is necessary to decide which authors should be permitted to return each paper in order to improve the overall income. In this talk, we discuss some of the complexities of the REF optimisation software.

The meeting will be held at University of Leeds, Room 9.27, Level 9, Worsley Building, 3:30pm until 5pm, Refreshments from 3pm