How Do We Do Science Mapping?
Though we do not specialize in epidemic research, we systematically search scientific literature on a given topic and visualize the connections between publications. In these networks, called science maps, we distinguish clusters that represent meaningful research fields, and describe them in the reviews. Here we explain the details of our procedure.
How can people outside epidemiology like ourselves reliably review the specialized literature? We have several nodes in our safety net.
1. Selected sources
After defining the topic of analysis, we turn to databases that contain information about scientific publications – Web of Science or Scopus. These bibliographic databases include only handpicked journals (Martín-Martín et al. 2018), which precludes low-quality research from appearing there. (True, many good publications are not indexed there as well, especially those not in English. Yet since reviewing implies reading the literature, lifting this limitation would bring little to the analysis.) We have full access, under the university subscription, to the indexed information, including the title, abstract, source, date, and list of references for each publication.
2. Searching skills
Working with bibliographic databases means speaking their language: it’s more than even advanced Google searching. Necessary terms should appear in adequate proximity to each other in the texts of abstracts; the diversity of concepts addressing the same research issue should be taken into account. We have learned, in practice, to craft search queries that result in the most relevant publications. In each of the Science map posts, we include the exact query that was used to retrieve the data.
3. Specialized software
After exporting the search results, we use the visualization software to process the bibliographic data and build the maps themselves: VOSviewer and CitNetExplorer (van Eck and Waltman 2017). In particular, the programs parse the reference lists of publications, which allows for analysis of indirect connections between the papers, and use an original clustering algorithm that guarantees good split of the literature into subsets (Traag, Waltman, and Eck 2019). The software was developed in the Dutch Centre for Science and Technology Studies – one of the leading organizations in the world working with scientific research data and methods of analyzing it.
4. Suitable strategy
Revealing non-direct connections between papers – when two papers share some references, or when two papers are addressed together in the same third one – increases the accuracy of mapping a science field (Boyack and Klavans 2010). The described approaches are called bibliographic coupling and co-citation, respectively. Substantially, if such a connection between papers is present, is it more likely that they discuss similar issues and belong to the same research subfield. Though bibliographic coupling and co-citation result in different types of networks, these networks quite similarly represent the research field (Yan and Ding 2012). Mostly we will be working with bibliographic coupling maps in VOSviewer, as technical affordances are more diverse for this kind of map. Occasionally we will use direct citation analysis in CitNetExplorer to explore citation trees within the literature and identify the figureheads of research traditions.
5. Sensible suggestions
Not only do we describe the structure of the research fields, but we also compose a list of recommendations on what to read if you are particularly interested in the topic. The lists of recommended literature in each Science map post consist of the most cited research. We also try to account for different aspects of the topic and balance the works in terms of the year of publication.
Still, our suggestions can be improved by yours – share with us your ideas on the possible topic for reviews, or approach to reviewing, or whatever it is – we will gladly consider any feedback. You can find our emails in the tab ‘Contributors’, or use the special form in the tab ‘Leave feedback’. Thank you and stay tuned!