Mapping impact pathways: improving our understanding of what mechanisms work in research translation

The successful translation of research knowledge to societal impact is a topic that has attracted the interest of governments around the world for a number of years. In 2014 the UK government introduced an ‘impact’ element to university assessment in the Research Excellence Framework (REF) to incentivise academics and their institutions to increase their contribution to society. Although there was a lot of push back at the time, in effect the government was saying: ‘demonstrate to us that about 10% of the research activity funded over the past 20 years in your university has made a contribution to society, and if you can we will give you more funding’. After the last two cycles of REF (2014 and 2021), the research team were involved in large-scale analysis of the impact case studies (ICS) commissioned by UKRI and Research England. Text mining approaches were used to identify impact topics based on the common use of keyword phrases in Section 4 (details of the impact) of the ICS. In combination with classification of the research disciplines (fields of research) that underpinned the research, it was possible to show the flow of impact from disciplines, through REF ‘units of assessment’, into impact topics. In both assessments the outcome of such analysis shows a multitude of impact pathways across all academic disciplines, leading to the conclusion that impact is extremely varied and difficult to categorise. This outcome was also observed in another study conducted by the project team on the results of the Hong Kong Research Assessment Exercise 2020. The current project aims to build on these analyses by investigating pathways to impact, based on a combined analysis of REF 2014 and 2021 ICS. The objective is to map common and unique pathways and examine their characteristics. Quantitative text mining will (i) map impact pathways from the underpinning research to impact topics for the combined dataset of 13,158 ICS, (ii) identify and describe common and unique pathways within the dataset, and (iii) identify any differentiating characteristics in the available metadata. Informed by the quantitative analysis, a series of ‘deep dives’ of common pathways will then be undertaken. The qualitative analysis will focus on cataloguing the types of mechanisms that have facilitated knowledge translation, and the beneficiaries of the research. A further text mining stage will apply this learning back to the whole dataset, seeking to automatically detect mechanisms that led to impact, along with beneficiary groups. This will result in a ‘data dictionary’ of impact mechanisms and beneficiaries, that can be applied to future text analysis of ICS. The project will contribute to our understanding of research impact and its beneficiaries, as well as having a number of real-world benefits. It will generate actionable insights for policymakers, universities and funders on effective approaches to support research translation. For example, it is likely that a particular mechanism is more frequent in one discipline than another, indicating the possible effectiveness of more targeted approaches to research support (including through impact acceleration accounts, technology transfer offices, etc). It will also generate guidance and toolkits for researchers on how to maximise the impact of their own research.