Pro-ana and pro-mia social networks. The promises of qualitatively-informed agent-based modeling
Tubaro, Paola and Casilli, Antonio A. (2010) Pro-ana and pro-mia social networks. The promises of qualitatively-informed agent-based modeling. In: 6th UK Social Networks Conference, 14 - 16 April 2010, University of Manchester. (Unpublished)
tubarocasilli6uksnc_web.pdf - Presentation
Restricted to Repository staff only
Mixed-method models are increasingly adopted in the social sciences (Lincoln, 2010). In this paper, we take a methodological stance by discussing the benefits of combining qualitative research (in-depth interviews, participant observations and case studies) with agent-based computer simulation in order to analyse social networks, in cases in which detailed micro-behavioural information is required and/or with hidden and sensitive populations.
The paper is part of a larger research project on the recent rise of a highly controversial Internet movement advocating anorexia nervosa and bulimia nervosa (“pro-ana” and “pro-mia”, in web jargon), in the form of websites, blogs and forums. In an effort to gain insight into the social determinants of this subculture, their implications for the health and nutrition of participants, and possible policy responses, we are conducting a comprehensive comparison of the online and offline personal networks of individuals with eating disorders.
The data in which we are interested allows limited use of quantitative methods. Large-scale surveys, longitudinal studies, or even mapping of complete networks can hardly be conducted with a relatively small, partly hidden, and very vulnerable population. To overcome these constraints, we have devised a general analytical framework relying heavily on qualitative research carried out on a smaller, purposive (i.e. not statistically representative) sample of subjects. The method consists of using this type of empirical data to inform agent-based computer simulation (Tubaro & Casilli, 2010). While qualitative analysis and models have often been considered as mutually exclusive, we claim that bringing together their respective strengths can contribute to a better understanding of social phenomena that would be difficult to study otherwise - notably the pro ana-mia lifestyle and computer mediated communication in the related “cloaked” websites (Daniels, 2009).
In particular, we illustrate how qualitative research can provide present-day agent-based models with a sound and proper insight as to the behaviour and motivations of social actors –an insight that quantitative data do not always provide, at least at micro level and for hidden and sensitive populations. Conversely, agent-based modelling and simulation offer valuable tools for: a) performing “thought experiments” to check the consistency of social scientists’ theoretical frameworks and to precisely define their domains of applicability; b) replicating and generalizing findings; c) providing a basis for cross-disciplinary validation of results.
We detail the epistemic and methodological structure that allows articulation between qualitative analysis and agent-based modelling to occur, insisting on the possibility for qualitative data to intervene at different stages of the modelling process and on the need for a two-way feedback, from data to model and from model to data.
Finally, we conclude with a discussion of potential challenges, both in relation to our particular research project (with its twofold online/offline perspective and its specific target population) and from a more general methodological viewpoint
|Item Type:||Conference or Conference Paper (Paper)|
|Uncontrolled Keywords:||mixed methods, qualitative research, agent-based models, social simulation, online social networks, personal networks, eating disorders, “pro-ana” and “pro-mia” websites|
|Subjects:||H Social Sciences > HM Sociology|
|Faculty / Department / Research Groups:||Faculty of Business > Centre for Business Network Analysis
Faculty of Business > Department of International Business & Economics
|Last Modified:||14 Oct 2016 09:13|
Actions (login required)
Downloads per month over past year