Research Impact conference: an update on REF

Report by Christine Geraghty
Chairperson, MeCCSA

Research Impact conference

On behalf of MeCCSA, I attended an event on research impact organised by Academy of Social Sciences on 16th July 2009, including a speech by Graeme Rosenberg, the the REF project manager for HEFCE .

HEFCE will produce consultation document in September with three month period of consultation

Proposed census date will be December 2012 with review taking place in 2013 and funding from 2014. A pilot study on impact will run in Autumn 2009 alongside the consultation.

REF will be based on three categories – outputs, impact and environment. Outputs will still carry highest weighting but impact should be at least 20 and possibly 30%. Would be best if there was as much consistency as possible across panels, including weighting.

Outputs will be assessed through expert review on originality, rigour and significance. Bibliometrics not robust enough to be used except on panels where their value is already established ie in some sciences.

Non-academic impact will be broadly defined and not linear ie not based on proof that this research caused that outcome. Want to assess impact not measure it. Assessment will look at reach (eg how may people felt it) and depth (how profound was the effect). Impact to be assessed through narrative evidence based particularly on case studies of specific examples and linked to high quality research. Case studies may be based on different outputs/staff than those submitted above and can relate to longer timeframe than 2008-2012. Number of case studies to be submitted will link to number of staff in submission. Impact will be assessed at level of department or research group rather than individuals or particular outputs. There will be a pro forma for submissions and examples of indicators. Users of research will sit on sub-panels/panels and will be particularly important for the assessment of impact.

Nothing much said about environment

Seeking to rationalise sub-panels in terms of volume of work and have fewer panels.

Users are not just government or big corporations but could be smaller, less powerful users. Recognise difficulty in recruiting these people and they could be exempt from looking at the outputs more generally.

Still looking for ways in which to reduce the burden of process now that bibliometrics won’t do that.

The full presentation can be downloaded from here