Agency Network Analysis (ANA) maps a given actor’s arena of social action in the context of a problem space articulated by that actor. This mapping is done via a multi-part interview process that elicits that actor’s social network map and cognitive map.
You can download the AgNeS app to run Agency Network Analysis, hosted on the UNAM LANCIS website. The software files are also available on the GitHub repository.
ANA was developed for researchers to better understand the agency of specific actors in a socio-ecological system (SES) prior to the implementation of a Transformation Lab (T-lab). It was developed by J. Mario Siqueiros-Garcia and Lakshmi Charli-Joseph of the National Autonomous University of Mexico (UNAM) in collaboration with the North American Hub of the PATHWAYS Network (part of the Transformations to Sustainability programme).
The T-lab, which comprises workshops and research conducted over a 3-year period starting in 2016, was designed to approach a ‘wicked’ sustainability problem through transforming individual and collective agency within the SES (see this blogpost for details). It was therefore important to understand the kind of agency that would be brought into the action arena of the T-Lab.
ANA combines social network mapping with cognitive mapping to better understand an actor’s agency in a socio-ecological system. In this context, social network mapping is used to identify who a given interviewee collaborates with the most, and how they collaborate in practice. This process creates a depiction of an actor’s ‘social action arena’ to understand their social capital (as depicted through the social network) and the interviewee’s role relative to the system.
We used the tool ‘ego-nets’, as it is known in social network analysis (Crossley et al. 2015), and complemented this tool with what we call ‘action-nets’, a tool we developed for this particular project.
In the ‘ego-net’ tool, the relationship of the ‘ego’ (the interviewee) with ‘alters’ (collaborators in the problem space) are depicted in terms of their positions in concentric circles, with the ego at the centre. In the ‘action-net’ tool, the relationships of actors are depicted in terms of the types of activities they collaborate on together, thus representing forms of agency within the system.
Following this, cognitive mapping is used to understand how actors perceive the problem space (in our case, this was the problem of urbanization and wetland degradation in Xochimilco). The problem space is elicited in terms of the causal relationships of elements and variables that shape the system.
To ensure that the interviewees situate themselves within the system, rather than as external observers, they are requested to identify where (in relation to which variables) they view their current capacity to intervene in the problem, and therefore feel empowered to act.
Ultimately ANA enables a mapping of the social action arena of a group of an actor within their perceived problem space by situating the relative position and importance (centrality, betweenness-centrality and clustering coefficient) of the alters (collaborators) of an ego (the interviewee). This makes it possible to work with the actors to help them identify their own space of action – in other words, where they feel they have agency in the system: over what elements and through what networks.
Broadening out and opening up
The ANA methodology can help the facilitators of a T-lab to ensure that it includes participants with diverse system framings, roles, and capacities. If one of the goals of a collective process is to challenge the way an actor frames a problem, it is important to ensure that an actor’s own frame will be challenged by those of others. Thus, a plural, rather than singular, set of perspectives is held by the collective.
Additionally, if a T-lab is aiming to transform individual and collective agency within an SES, the ANA method functions not only as an important way of establishing a baseline for understanding an actor’s agency before the T-lab, but also a way to ‘broaden out’ a participant’s views before they participate in the T-lab. The interview process is a provocative one that requires an actor to reflect deeply and thoroughly on the system and their agency within it with renewed vigour.
Fit and limits
The ANA methodology is based on the articulation of three different tools: Ego-nets, Action-nets and Cognitive maps. The ‘articulation’ of these tools means not only adding them up, but linking them conceptually in order to describe an actor’s ‘agency profile’ from the perspective of the actor themselves.
One of the strengths of ANA is that its three tools are framed in the same networks language, which means they can be connected in a way that allows the creation of an integrated narrative about the actor’s agency profile. The narrative is structured in a sequential way as follows:
1) who the actor collaborates with (ego-net),
2) doing what (action-net) and
3) where collaborative actions impact on the system (cognitive map).
This method allows identification of the different routes from 1 to 3, detecting possible action pathways in which the actor can get engaged to act on the system. Once the routes are identified, they can be questioned by the interviewee.
In future developments of this method, we expect to be able to create indicators of how the system would change through the activation of the different pathways. ANA was developed to be used as a single, integrated method – but the outcome of every tool can be analysed independently using standard social network analysis.
From a more pragmatic perspective, applying these tools is very intuitive for both the interviewer and the interviewee. As they are applied in the context of an interview, if the interview is recorded, rich contextual information can be collected and supported by the network analysis.
Finding the limitations of ANA will require further exploration of the consistency, validity and reproducibility of the method. The application of ANA is time-consuming (up to three hours) and it may require more than one session per interviewee. Data processing (mostly extracting and cleaning data) from its raw record may also be time consuming. Those using this method will also need to learn how to use some software for network analysis; any social or complex networks analysis software will suffice, eg Cytoscape, Pajek, Gephi, Netdraw, or Mental Modeler (the latter being useful in generating Fuzzy Cognitive Maps).
Material for this vignette was developed by Lakshmi Charli-Joseph, Mario Siquieros and Hallie Eakin, with assistance from Rebecca Shelton, David Manuel-Navarrete and Beatriz Ruizpalacios. The AgNes software package was formalised and documented by Rodrigo Garcia Herrera.
To use the AgNes app, visit the download page or view the documents on GitHub.
- Crossley, N. Bellotti, E. Edwards, G. Everett, MG. Koskinen, J. and Tranmer, M. (2015) Social Network Analysis for Ego-Nets, SAGE. London: UK