Title

Finding the common threads - describing and predicting linkages impact

Loading...

Media is loading
 

Document Type

Video

Abstract

Researchfish, Interfolio, United Kingdom

Background: All parts of the research community have an interest in understanding research impact whether that is around the pathways to impact, processes around impact, methods for measurement, describing impact, etc.

The UK Research Excellence Framework (REF) requires Higher Education Institutions (HEIs) to make submissions, including Impact Case Studies (ICS), to demonstrate their impact. This requires HEIs to review and synthesize copious data, including the contents of Current Research Information Systems. In 2014 the estimated cost of producing ICS for the REF was £55M. (https://www.technopolis-group.com/wp-content/uploads/2020/02/REF-Accountability-Review-Costs-benefits-and-burden.pdf)

A recent study (https://doi.org/10.12688/f1000research.74374.2) explored the relationship between research funding awards and research impact using the ICS submitted to REF2014 as a proxy for impact.

This paper develops a method to link awards to ICS and predicts which awards would underpin ICS i.e. are thought to be "high impact". The case of REF2021 is used but the method helps predict "high impact" projects, meaning that this should be generalizable to other assessment exercises, or other research management activities focussed on impact.


Methods:
Linking - This paper uses REF2021 Data (https://results2021.ref.ac.uk/impact), funding acknowledgements, and information in Researchfish (an online platform used to collect information on the outputs, outcomes, and impacts of research) as part of a multistep approach to develop links between awards, publications, and ICS.

Predicting – This paper uses information collected in Researchfish and multivariate probit models to predict which awards would be included in ICS as part of REF2021.

Results:

The previous approach using REF2014 data was significant in predicting which awards were linked to ICS, despite a number of issues with the dataset (particularly time based coverage of data).

Preliminarily results of the REF2021 analysis indicate that the final linkage will likely be higher and more accurate than REF2014. This should also reduce error and improve the performance of the predictive modelling.

Significance:
If successful at predicting which awards will be included in a ICS this could provide the basis for a tool to help an HEI identify high impact research for ICS in impact exercises, improving efficiency and reducing costs.

Publication Date

11-2022

This document is currently not available here.

Share

COinS