By Leonhard Späth, Rea Pärli and the RUNRES project team
Can we observe in a more analytical way how transdisciplinarity “happens”? How useful is social network analysis in transdisciplinary work, especially for uncovering the role of relationship structures? How can transdisciplinary concepts be used to map connections between those involved in transdisciplinary research?
A very brief introduction to social network analysis
Social network analysis is the study of connections between different people or any other social entity involved in the topic under investigation (referred to as actors), as well as the patterns of those connections and the distribution of the ties among actors.
There are many ways to conduct a social network analysis. The first step is often to identify the relevant actors. The second step is to find out about a specific relationship with the other identified actors. This could involve asking the actors with whom they communicate or with whom they work or any other form of connection. The data are generally collected through surveys, interviews or through secondary sources such as data analysis. The data are then used to illustrate and analyze the networks using various tools.
Operationalizing key transdisciplinary concepts for social network analysis
For transdisciplinary research to be studied using social network analysis, key concepts should be formulated as questions about interactions between actors that can be answered with a “yes” or “no” response. We illustrate how this can be done with three concepts relevant to transdisciplinary processes: three types of knowledge, four levels of involvement and three rationales for involvement. Each actor in the project is asked each question in relation to each other actor in the project.
1. Three types of knowledge:
- System knowledge: “I provide knowledge such as scientific knowledge or experiences to him/her about topics and innovations in [the project] domains.”
- Target knowledge: “I provide him/her knowledge about what are desirable outcomes in [the project].”
- Transformation knowledge: “I provide him/her knowledge on how to achieve the goals we have set in [the project].”
2. Four levels of involvement:
- Information: “I make him/her aware about topics and innovations in [the project] domains.”
- Consultation: “I ask her/him opinions or information on [the project] issues.”
- Collaboration: “We jointly address issues within [the project] together.”
- Delegation: “I delegate to him/her tasks related to [the project].”
3. Three rationales for involvement:
- Substantive rationale: “I provide him/her information relevant to [the project] that s/he does not have.”
- Instrumental rationale: “I need him/her on board to collectively achieve the goals set by [the project].”
- Normative rationale: “I interact with him/her because I think it is fair to have him/her in [the project].”
Studying interactions in a transdisciplinary project
An example of what social network analysis can contribute to transdisciplinary research is shown in the figure below and the corresponding discussion. The example comes from a project in South Africa, where we are implementing technologies to re-circulate nutrients from organic and human waste from urban areas back to agriculture in rural peripheral areas. We used social network analysis to provide insights into:
- what the different actors exchange, in our case what type of knowledge is exchanged.
- how the actors exchange, in our case what type of involvement they use.
- why the actors interact, in our case whether it is for substantive, instrumental, or normative reasons.
The top section of the figure (blue dots) shows how the actors exchange the different types of knowledge. The bottom section (red dots) shows the reasons for the different actors to exchange knowledge. Each blue or red dot (called a node) represents an actor, with the size of the node indicating its centrality, which represents how strongly interconnected the person is in the network (the larger the node the more interconnected the person is). N is the number of actors and d is the density, calculated as the share of total ties in the network divided by the share of possible ties to all other people, which represents how interconnected the different actors are.
In this example, we can observe that the network of system knowledge includes more actors and has a denser network structure than the other two knowledge types. This suggests that in this transdisciplinary project actors share system knowledge more than target and transformation knowledge. Such information could be used to modify the process by actively involving more actors, especially those previously excluded, in the production and sharing of target and transformation knowledge.
When examining the bottom third of the figure (the rationales or “why”), we observe that substantive interactions have the densest network, followed by instrumental interactions. Purely normative interactions happen between comparatively few actors; only 25 of 69 possible actors are part of this network. This suggests the main rationale for interaction, substantive, is because actors need information from other actors and not because they feel that it is fair to share knowledge with them.
Social network analysis also makes it possible to study how different networks are connected and potentially influence each other. This is shown in the middle section of the figure, which illustrates which knowledge type (“what is shared”) is connected to which rationale (“why is it shared”). The thickness of the arrows and the corresponding numbers represent the strength of the correlation ranging from zero (not correlated at all), to one (completely correlated).
The strongest correlation (0.68) is between transformation knowledge and substantive rationale. The weakest correlation (0.12) is between transformation knowledge and normative rationale. Although the results must be interpreted with caution, this may suggest that actors came together in this transdisciplinary process to achieve their own ends, rather than as an end for itself.
As this example shows, using social network analysis to describe and analyze transdisciplinary concepts as well as their interactions can inform and stimulate a reflective process on the way that different actors interact and are involved in a project. This can be useful for examining the dynamics of a project, especially from a participatory perspective. It may also indicate when modifications to a transdisciplinary process should be considered.
If you have used social network analysis in transdisciplinary research, what have you investigated and observed? If you are a transdisciplinary researcher who has not used social network analysis, are there other questions about connections that would be useful to study?
To find out more:
This research was undertaken as part of the “RUNRES – Establishing a nutrient-based circular economy to improve city region food system resilience” project, conducted in the Democratic Republic of the Congo, Ethiopia, Rwanda and South Africa and funded by the Swiss Agency for Development and Cooperation. (Online): https://runres.ethz.ch/
This research will be presented at the 2021 International Transdisciplinarity Conference to be held online from September 13–17, 2021. (Online): https://transdisciplinarity.ch/de/veranstaltungen/itd-conferences/itd-conference-2021/
Biography: Leonhard Späth PhD is a postdoctoral researcher at the Department of Environmental Systems Science at ETH Zurich, Switzerland. He focuses on three main challenges around organic waste recovery for agriculture in East Africa: integrating different stakeholder-perspectives through transdisciplinary methods, structuring decision-making processes for more inclusive decisions, and monitoring sustainable development, with a focus on the Sustainable Development Goals.
Biography: Rea Pärli is a PhD candidate at the Department of Environmental Systems Science at ETH Zurich, Switzerland. She has a background in environmental systems and policy. In her current work she explores how transdisciplinary research projects work, what results they produce and how they contribute to sustainable development. She is interested in what project designs and processes support transdisciplinary research and how these factors are influenced by each other and external factors.
Participants: RUNRES (Rural-Urban Nexus Research) project team: Abayneh Feyso, Abebe Arba, Behailu Merdekios, and Kinfe Kassa (Arba Minch University, Ethiopia); Benjamin Wilde, Johan Six (project PI), Mélanie Surchat, Leonhard Späth, Pius Krütli, and Rea Pärli (ETH Zurich, Switzerland); Haruna Sekabira (until mid-2021), Kokou Kintché, Marc Schut (until end of 2020), Matieyedou Konlambigue, Moustapha Byamungu, Murat Sartas, and Speciose Kantengwa (International Institute for Tropical Agriculture, Democratic Republic of the Congo and Rwanda); Alfred Odindo, Ndoda Zondo, Samuel Getahun, Sharon Migeri, Simon Gwara, and William Musazura (University of KwaZulu-Natal, South Africa). Not everyone is shown in the photo below, which can be expanded by clicking on it.
8 thoughts on “How can social network analysis benefit transdisciplinary research?”
Dear authors, this is a very interesting and timely approach of SNA in transdisciplinary research. I am curious about the practicalities and actual impact of such research for your practitioners.
First, from my own experiences of doing SNA in transdisciplinary research, I found it extremely time consuming for practitioners to answer all the SNA questions. How did you do this in an efficient and engaging way? Do you have any recommendations here?
Second, when we presented our SNA results to our practitioners, it was very difficult for me to derive really beneficial and useful insights for the practitioners they they could immediately use. So how did you provide your SNA results in a beneficial und really useful form to your practitioners? Did your practitioners take up your SNA results and worked with them?
Third, was your SNA co-designed and co-developed? Or why did you decide to do a SNA?
I would love to hear your reflections on a this. Thank you!
Thank you for your comments and questions! We used an interview format to collect the SNA data. This means that somebody of the team asked the participants about their relationships rather than them filling out the survey. While this method is more time consuming we assume it slightly more engaging. However, we are really interested in exploring other ways of data collection which are more engaging and maybe even fun (e.g. gamification).
So far, we have used SNA mainly as a tool to explore a TD process but not really as a TD tool itself. We will present and discuss our detailed results with the stakeholders in the next weeks. In other projects using SNA, we found that the results were useful especially for documenting and planning the stakeholder engagement process and activities. The project leaders (from research and practice) used the different maps to strategically address stakeholder groups which have not really been included in the discussion before. We expect to have similar benefits in our project, and we can let you know how this turns out.
Our SNA was partially co-designed. For the identification of relevant stakeholders for the SNA, we used a purposeful sampling approach, meaning project members and other content experts in the field jointly brainstormed to identify the relevant stakeholders. Further, during the SNA interviews all respondents had the opportunity to add additional names. However, we did not include the stakeholders in the decision of using SNA as this decision was mainly based on the need of having a (rather disciplinary) tool to observe and map transdisciplinary research projects through concepts that are already established in the TD literature.
I really like the approach, which acknowledges the different types of involvement, types of knowledge and rationales in the interactions of transdisciplinary projects. Working previously with network analyses for project evaluation, though (see DARE project: http://www.sussex.ac.uk/spru/research/dare/index.html), I experienced the difficulty of making sense out of the SNA measures (you get this measure — so what?). I also wonder about the difficulty of separating between types of knowledge (how can you provide transformation knowledge without having target knowledge) and even more between the different rationales. Is it not often the case that stakeholder involvement will often have overlapping normative, instrumental and substantive rationales?
Thank you Ismael for your comment. We indeed have these difficulties in the approach we take, the “so what?”, hence our classification of the different concepts as “what”, “how” and “why”. However, we aim to put these SNA results in a broader context of evaluation of transdisciplinary processes, which is mainly done through workshops. On top of this, we plan as a subsequent research to investigate the validity of what we measure through qualitative interviews. Regarding the overlaps, yes, they already exist in the original descriptions of the concepts we use, and we can also see that in our results. The participants always had the option to select multiple concepts, meaning multiple types of knowledge and multiple rationales. We are now also running statistical network models to identify the strength of the different connections between the concepts. Thank you for sharing the DARE-perspective. We think that it provides an insightful way to evaluate collaboration processes. We will check further into this and we believe our and your different variables may complete quite nicely.
Excellent piece of work. I really like the way you have carefully defined the different types of relationship and then looked at network measures for the association between relationships. I wasn’t quite clear on what you are correlating though. Is this something like Jaccard index (so presence/absence of the edge), or are you measuring correlation between centralities of the same node in different networks (or something else)?
Thank you for this comment and your question! We correlate the tie-structure of the networks. In a nutshell this means we are looking at the similarity of different networks in terms of the ties the different edges form (similarity measure for graphs). As a next step of the data analysis, we are now running Exponential Random Graph Models (linear regressions for networks) to find the effect size of the correlations.
This is going to be a great resource to explain SNA to folks who do not use it regularly. And love the: three types of knowledge, four levels of involvement, and three rationales for involvement. Thank you for writing this blog post!
Thank you Hannah! We also hope that SNA will be used by transdiciplinarity practitioners as a useful ingredient for a better understanding of transdisciplinary processes.