RAE Review: Interim feedback from facilitated workshops Jonathan Grant and Steven Wooding
Mar 31, 2015
RAE Review:
Interim feedback from facilitated workshops
Jonathan Grant and Steven Wooding
Plan
Introduction Who we are and what we do Where we went and who we spoke to What we did and why we did it
Results from 6/9 workshops Task 1: What is quality research? How should it be assessed? Task 2: Assessing model systems Task 3: Building a better system Task 4: Implementation and implications
Some emerging observations Preference for expert review, but needs refining Increased transparency and clarity of process Tension between comparability and appropriateness Need structures to support change
Who we are and what do we do
RAND Europe
Independent not-for-profit public policy think-tank
‘Cousin’ of RAND; US based independent think-tank employing 1600 researchers
RAND Europe. Established in Leiden (NL) in 1992; Cambridge in 2001
In UK programmes include Transport, Information Society, Health and R&D Policy
Current projects include VFM study on government department research for NAO and on scientific mobility for WHO
Where we went and who we spoke to
Based on 6/9 Workshops
93 people
Positions
26 Administrators 38 Senior academics 13 Academics 8 Research Fellows 8 Unclassified
Fields 32 Medicine, science & engineering 16 Social Science 15 Arts and Humanities 29 Not research active 1 Unclassified
*Excludes Cambridge, Reading & Belfast (n=c50)
What we did and why we did it
Facilitated workshops
Provide framework for structured thinking
Captured outputs in a standard and comparable form
Allows comparison between mixed and like groups of people (e.g., administrators only vs. mix of HoDs, fellows, and research officers)
Purpose is to listen, not evaluate
Agenda
Task 1: What is quality research? How
should it be assessed?
Task 2: Assessing model systems
Task 3: Building a better system
Task 4: Implementation and
implications
Task 1: What is high quality research? How should it be assessed?
Purpose
Stimulate wide ranging thinking on the most important aspects marking out high quality research and research assessment systems
Task 1
Introductions
Identify 5 characteristics of high quality research
Identify 5 characteristics of research assessment system
Vote (5 votes for each; allocated as seen fit)
‘Top 10’ characteristics of high quality research
0 10 20 30 40
Innovative
Longevity
Value for Money
High Quality Publication
Academic Impact
Rigour
Peer Recognition
Defining Research Agenda
Original
International
‘Top 10’ characteristics of research assessment
systems
0 10 20 30 40 50 60 70
Resists Game Playing
Cost Effective
Advance
Not Burdensome
Rigour
Quality not Quantity
Comparable
Explicit
Flexible
Transparent
Task 2: Assessing models systems
Purpose
Evaluate the strengths and weaknesses of 4 ‘model systems’: Expert review; Algorithms/metrics; Self-assessment; and, Historical Ratings
Task 2
Split into 4 groups of c4-5
2 groups look at ‘good’ aspects of 2 systems each
2 groups look at ‘bad’ aspects of 2 systems each
Also identify questions that need answering
Algorithms / metrics
Selected questions How to recognise novelty? What do you count? How to ensure comparability?
‘Good features’ Transparent (4) Objective (3) Cheap (3) Simple (2)
‘Bad features’ Not suitable for all subjects (5) Spurious objectivity (3) Open to ‘game playing’ (2) Metrics are proxy measures (2)
Expert review
Questions Who are the experts and how are they selected? How do you build in an appeals mechanism How do you recognise innovation?
‘Good features’ Acceptable to community
(5) Based on specialised
knowledge (3)
‘Bad features’ Not comprehensive (3) Not transparent (3) Perceived bias (3) In consistent (2) Expensive (2)
Historical ratings
Questions How do you take account of changing performance? Who makes the judgement? (HEI or individual) How far back?
‘Good features’ Light touch (3) Cheap (2) Ability to plan (2)
‘Bad features’ Inhibits change (3) Low credibility (3) Perpetuates silos (2)
Self-assessment
Questions How would you police it? Who sets the goals? How do you penalise inflated results?
‘Good features’ Sensitive to discipline (2) Formative – considers self
(2) Ownership & Trust (2)
‘Bad features’ No cross discipline
comparability (3) Open to game playing (3) No confidence in system (2) Effort could be large (2)
Task 3: Building a better system
Purpose
Design ideal research assessment system
Task 3
Split in to 3 or 4 different groups of c4-7
Select ‘seed’ system from Task 2
Build on this using aspects of other system
Present back to plenary
Identifying a base system
Starting point
Expert review (16/18 breakout groups)
Self-assessment (1/18 breakout groups)
1 ‘failed’ (to reach a decision)
(remember based on 6/9 workshops)
Refining the expert system
Transparency and clarity of process
Establish rules at outset
Early (i.e., up to 5 years before assessment)
Don’t change rules during process
Provide feedback
Legal contract between FCs and HEIs (rules won’t change, in return for guarantee of no challenge)
Clarity of funding outcome
Refining the expert system
Longer time period between
assessments
Review every 8-10 years
‘Triggering’ mechanisms for interim review at 4-5 years.
Self declaration for new or emerging areas
Metrics for decline areas
Sampling (selective or random) with other departments
Refining the expert system
Broader panels
Mirror Research Councils/AHRB
Broad or supra-panels have broad freedom to establish discipline specific rules
Sub-panel operate to those rules, reporting to supra-panel
Aims to solve tension between comparability and appropriateness
Refining the expert system
Continuous rating scale
More grades
Summation of individuals ‘scores’
Possibly based on ranking
Continuous funding scale
No (or reduced) step changes in funding
Refining the expert system
Conceptual (non subject) AoAs
Based on research outputs
Formal sciences Theorems
Explanatory sciences Laws
Design sciences Technological rules
Human sciences Artifacts and knowledge
‘Users’ from relevant AoA
Refining the expert system
‘Lay‘ Panel members
Possible experience lay members (such as judges) as panel chair
Rolling review
One discipline every year
Transfer fees
Reward departments that nurture future high flyers
Return all staff
Task 4: Implementation and implications
Purpose
Examine system would be put into practice and evaluate repercussions
Task 4
Working in same groups with ‘devil’s advocate’
Identify 5 steps for implementation
Identify 5 things that could go wrong with the system
Identify 5 changes to the UK research as a result of the new research assessment system
Implication of change
Some changes to UK research:
Ensure research capacity is activated wherever it is found
Better support and funding for younger researchers
Good support for emerging areas of research
Equal opportunities are improved
Better recognition of interdisciplinary research
Funding directly following excellence
Less obsession with the RAE
Emerging observations
Preference for expert review, but needs refining
Increased transparency and clarity of process
Tension between comparability and
appropriateness
Need structures to support change
But ….
Excludes Cambridge, Reading and Belfast
Not analysed by discipline and or profession