Using a database with information on over 1,500 researchers, statistical analysis was recently undertaken to test the hypothesis that technical STEM subjects were more societally useful than social science and humanities (SSH) subjects. Paul Benneworth describes the research process and the findings which suggest SSH research is just as useful as STEM research. A less stereotypical understanding of impact is required to maximise investments and create benefits for society as a whole.
Anyone active in research cannot help but notice increased recent pressure from research funders to maximise its wider societal benefits. Ensuring public benefits in return for funding is clearly reasonable in a democratic society, but this increasing drive for impact has brought with it a rather undesirable set of policy assumptions, prejudices and stereotypes of which research creates public benefit.
These assumptions affect how policy decisions regarding research funding in general, and specifically the treatment of different disciplines. Recent Horizon2020 programme discussions perfectly exemplify this, purporting to focus research on ‘grand societal challenges’ whilst framing those challenges almost exclusively in technical terms. This has the effect of making social science and humanities (SSH) research seem peripheral, with no intrinsic societal value beyond helping other technical disciplines solve their own important problems.
This belief in science, technology, engineering and mathematics (STEM) disciplines’ superior societal benefits over SSH might well be true. But there is much evidence that SSH research is societally useful, including Jonathan Bate’s highly engaging edited collection on The public value of arts and humanities research and Alan Hughes et al. Hidden Connections report. So we can’t take as axiomatic that STEM is more societally useful than SSH: there is a prima facie case that this assumption demands further rigorous testing.
This is the background for research I undertook together with colleagues at CSIC-Ingenio in Spain, published earlier this month in Science and Public Policy [pdf]. Our argument was quite simple – if SSH research was less useful than STEM research, then you would expect there to be fewer people that found that research useful. In that case, you would also to see expect differences in the behaviours of SSH researchers in response to less user pressure. To try to understand these differences, we reviewed policy and scholarly literature to identify the “claims” that were made about differences between STEM and SSH.
Some kinds of differences would claim that STEM is more useful than SSH, for example that SSH has difficulties in giving concrete answers to user questions. But other kinds of difference suggest instead that SSH is equally useful but is different in the way it engages with its users to STEM research. STEM user interactions often take place within (more easily countable) contractual relationships, but the fact it is difficult to count SSH researchers’ non-contractual knowledge exchange relationships does not mean a priori that they are less useful.
On that basis, we designed a set of eight hypotheses, four corresponding to the claims that we found in the literature that STEM was more useful than SSH research (1-4), and five corresponding to claims that STEM was simply differently useful to SSH research (5-9).
We operationalized these variables to permit tested using the IMPACTO database, a survey of researchers at all levels in the Spanish Consejo Superior de Investigaciones Científicas (CSIC) organisation. CSIC funds Spanish public research laboratories in all fields, and the questionnaire covered their research profile and activities, their relationships with non-academic users, obstacles to engagement and engagement outcomes. The survey covered 37% of the 4200 CSIC researchers across all disciplines, with only agricultural sciences over-represented in the survey.
The precise mechanics of operationalization are described in the paper: for all hypotheses, a null hypothesis was constructed that STEM and SSH performed similarly, and tested for significant evidence for rejection. For all but H4, simple categorical data was tested using a U-test. For H1, for example, the variable of national orientation was constructed on the basis of the relative intensity of their reporting of working with different kinds of users in Spain and abroad in the previous 3 years, as a ratio of national: international users.
For H4, scientists were allocated an identity category along two dimensions, the extent to which they pursued excellence in research and impact respectively, following Stokes’ categorisation. This gives four categories, and nul hypothesis was tested that the distribution of identities would be the same in these categories for SSH and STEM researchers, tested for significance with a χ2 test. The results of those statistical tests are summarised in table 2.
The results are quite dramatic: for all the ‘differently useful’ variables, SSH researchers behave demonstrable differently to STEM, being more likely to engage in informal interaction (e.g. temporary placements, serving on committees or professional meetings), popularisation, and work with government and non-profit organisations. Conversely, STEM researchers are more likely to use formal interaction (contract research, spin-offs, patents, shared infrastructure) and much more likely to work with firms.
For the variables which suggest STEM is more useful than SSH, only one is statistically significant, an orientation towards engaging with national over international users. The other three all suggest that SSH researchers behave in ways that suggest their research is just as useful to society as STEM researchers. And contrary to Hughes, we found no difference in academics’ engagement identities between SSH and STEM researchers, and certainly no evidence SSH were more oriented than their STEM colleagues towards blue-skies, ‘ivory tower’ research. Even the first result could perhaps be interpreted that SSH researchers are more likely to ensure their research is used for national benefits than STEM researchers.
We appreciate that our research is exploratory, drawn from a single country, and testing an existing database rather than undertaken specifically for our hypotheses. Nevertheless, our research gives us no evidence to simply dismiss our case that the policy-makers’ are making a false assumption. Of course, we aren’t claiming that SSH is always as useful as STEM research, just that we have evidence to challenge this policy-makers’ a priori assumption.
A gut feeling might tell you that shiny spin-off companies are more valuable for society than academics working with community groups, but a gut-feeling is no basis for effective policy-making. A less stereotypical understanding of how researchers create impact is clearly urgently required to help policy-makers and researchers work together more effectively, and maximise these massive investments to create benefits for society as a whole.
Note: This article gives the views of the author, and not the position of the Impact of Social Science blog, nor of the London School of Economics. Please review our Comments Policy if you have any concerns on posting a comment below.
Image credit: Kiran Foster (CC-BY)
Why ‘societal’? Why not ‘social’? Is there a difference, or are multi-syllabled neologisms just inherently better?
It’s a technical distinction.The word societal is used to avoid confusion with the idea of ‘social’ which means people doing things together interactively, in the sense of social media. Societal research impacts usually come through ‘social’ processes (because it is the way ‘tacit’ knowledge is transferred), but if i write an article and you simply read it, then that is a societal impact without an underlying social learning process (which would involve you and i having a discussion, as we are now doing).