How would you define your discipline? A team at Loughborough has devised a formula that gets to the heart of the subject
How can a discipline be defined? In looking for the answer, we considered a listing of journals by discipline published in the Science Citation Index. Journals were ranked according to quantitative estimates of the impact that an article could be expected to generate. Some journals made appearances in more than one discipline and we used this to calculate the "relatedness" of disciplines to one another. The greater the number of shared journals between any two disciplines, the more closely related they are.
We mapped the limits of the domain over which individuals in a discipline published. The term "discipline" is misleading here. What is meant is the population of individuals working in departments sharing the same name. We conducted such a survey in our own department, the department of chemical engineering, first. The extent of the publication domain was considerable, and we divided it into 12 categories. Some comprised a number of SCI-coined disciplines - our definition "chemistry" includes the SCI ones of chemistry (analytical), chemistry (organic) and so on. We then evaluated the publication output of senior academics in top-rated chemical engineering departments.
Each item of an individual's output over a five-year period was computer-matched by journal against the journal listings in our 12 discipline categories. Scores were assigned by multiplying journal impact factors by the number of articles published in that journal and then by summing the products in each category. An individual thus became characterised by a system of 12 numbers. We plotted these data in the form of radar plots. This revealed "footprints". All categories of life were represented - odd and even-toed ungulates, reptiles and small mammals.
By averaging scores in each category we compared groups of individuals, for example, one department with another. The most interesting was the comparison of the UK with the US. The US footprint covered an area 300 per cent greater and differed in morphology from the UK footprint. The former revealed incursions into biology, chemistry and medicine. We extended the analysis; the scores within the categories can be thought of as representing quantitative characters of what numerical taxonomists like to refer to as "operational taxonomic units" (OTUs). We calculated similarity coefficients between OTUs. In this case, these can be individuals, departments or countries. This yielded "phenetic information".
When applied to individuals in a department, the technique did not reveal anything unexpected. But what in research terms are departments anyway? Paraphrasing a former prime minister, "there are (only) individual men and women and there are families". One certainly sees clusters of individuals working in closely related fields, and some spectacular "outliers". Surprises were revealed when departments were compared with each other or with "type" - essentially the publication profile based on impact factor listing by discipline. Some prestigious departments in the US were mapping at considerable distances from type. Chemical engineering in the US has re-invented itself. As jobs in the traditional sectors of the chemical industry disappear, UK chemical engineering is groping for a way forward. As for the research assessment exercise, rigid definitions of discipline might hamper attempts for chemical engineering to reposition itself. This research was conducted with Klaus Hellgardt and information scientist Charles Oppenheim.
Gilbert Shama, department of chemical engineering, Loughborough University.