Mapping societal needs into research trajectories in agriculture and health: Results from three exploratory studies

Abstract

Scientific advance is unevenly distributed across sectors and social needs (Gibbons et al. 1994). Reasons for this include changing needs themselves, market and public organisation failures, inequalities of power and voice, intractable ‘wicked problems’, and scientific and technological path dependencies (Bozeman, Slade, and Hirsch 2011; Nelson 2011). For instance, technological developments in agriculture privilege specific forms of productivity at the expense of sustainability and tend to neglect local needs (Dalrymple 2006); policy and research attention is biased towards diseases of the industrialised rich, at the expense of the poor in developing countries (Evans, Shim, and Ioannidis 2014). Most conventional bibliometric studies have focused on the mapping of knowledge production side in terms of mapping research areas and measuring scientific influence, in a way that implicitly assumed an "internalist" view of science (see Rotolo et al. (2014) for a different view). In contrast, grand challenges such as agriculture (food and nutrition) and health, require investigating the articulation of societal needs. In the interest of a broader impact assessment, there is a demand for tools to study: (a) how societal needs translate into public research outputs; (b) why particular societal needs are prioritised significantly more than others; (c) why some are absent in public science. In this paper we present exploratory investigations comparing distributions of research topics and of societal priorities, addressing three issues that are relevant to broader impact assessments. We illustrate potential approaches using semantic networks and clusters based on publications' abstracts and keywords to map existing science supply.

First, addressing rice research, we compare the evolution of the research trajectory for various countries and organisations (e.g. IRRI), and related to different societal demands and contexts – e.g. rice exports in Thailand, crop yields in India, and privatised seed technologies in the US. We use co-word analysis to map cluster of topics and their evolution over time. We find that countries’ and organsiations’ trajectories differ significantly, reflecting different national innovation systems, capabilities, and priorities. However, the incentives shaping these trajectories are not always correlated with changes in societal needs. Second, addressing the grand challenge of obesity (a health problem with enormous social costs), we use question records in the French and EU parliaments as an instance of social demand (captured by decision makers). We use topic modelling to reconstruct thematic structures in both parliamentary data and publications (abstracts). We compare them with publication maps to explore (mis)alignments between societal concerns and scientific outputs. We find that research is more concerned about the biophysical mechanisms leading to obesity, whereas political questions focus on the social mechanisms that favour obesity. Third, focussing on published research on rice we address the issue that results may be strongly biased by the use of different databases. Using different clustering methods we compare the distribution of topics, for different countries, in publications from different databases: WoS, Scopus and CABI. We show that bibliometric exercise that use dominant databases may have a negative effect on policies relevant to important social issues, particularly in developing countries.