Collecting and analyzing qualitative data for system dynamics: methods and models

Abstract

System dynamics depends heavily upon quantitative data to generate feedback models. Qualitative data and their analysis also have a central role to play at all levels of the modeling process. Although the classic literature on system dynamics strongly supports this argument, the protocols to incorporate this information during the modeling process are not detailed by the most influential authors. Data-gathering techniques such as interviews and focus groups, and qualitative data analysis techniques such as grounded theory methodology and ethnographic decision models could have a strong, critical role in rigorous system dynamics efforts. This article describes some of the main qualitative, social science techniques and explores their suitability in the different stages of the modeling process. Additionally, the authors argue that the techniques described in the paper could contribute to the understanding of the modeling process, facilitate communication among modelers and clients, and set up a methodological framework to promote constructive discussion around the merits of qualitative versus quantitative modeling. Copyright © 2003 John Wiley & Sons, Ltd.Syst. Dyn. Rev. 19, 271-296 (2003) System dynamics is a powerful tool in the creation of feedback theories. Since its beginnings, the founders of the field have developed a series of guidelines for the model building process (Randers 1980;Richardson and Pugh 1981;Roberts et al. 1983;Wolstenholme 1990;Sterman 2000) and a series of tests to build confidence in the models created (Forrester and Senge 1980;Sterman 2000). As depicted by the classical literature, the development of system dynamics models is an iterative process. Each iteration results in a better and more robust model. Although system dynamics models are mathematical representations of problems and policy alternatives, it is recognized that most of the information available to the modeler is not numerical in nature, but qualitative. For example, while describing the information sources for the model building process, Forrester (1994) suggested that these qualitative data reside in the actors' heads (mental database) and in the form of written text (written database). Moreover, he recognized that the most important source, both in quantity and significance for the modeler, is the mental database (Figure 1): 1 As suggested by the figure, the amount of available information declines, probably by many orders of magnitude, in going from mental to written information and again by

The uses of qualitative data in multimethodology: Developing causal loop diagrams during the coding process

“…It is broadened by reference to similar methods used by (Burchill & Fine, 1997;Morrison, 2003;Repenning & Sterman, 2002) and exhortations from the system dynamics community for integration with qualitative data collection and analysis (Kopainsky & Luna-Reyes, 2008;Luna-Reyes & Andersen, 2003).…”

“…There is also strong advocacy for this type of approach from Luna-Reyes (Kopainsky & Luna-Reyes, 2008;Luna-Reyes & Andersen, 2003) and more recently from (Kim & Andersen, 2012 (Strauss & Corbin, 1998). They do however note the close matching between the linking of concepts in grounded theory and causal links in CLD.…”

“…We also observe that there is a direct link between grounded theory and systems modelling. The processes of grounded theory therefore provide a means for grounding systems modelling, especially structural modelling approaches such as CLDs and system dynamics (Kopainsky & Luna-Reyes, 2008;Luna-Reyes & Andersen, 2003;Sterman, 2000;Yearworth, 2010).…”

Exploring stability and change in transport systems: combining Delphi and system dynamics approaches

“…The sources of data for these models vary considerably, from direct empirical research and reviews of reports and literature, to consultation with stakeholders and subject matter experts (Luna-Reyes and Andersen 2003).…”

“…Eskinasi et al (2009), for example, involved stakeholders and policymakers to develop an SD model in an urban transport context and found that the process yielded counterintuitive insights which may have been otherwise missed, and helped stakeholders to work through contentious issues. These collaborative approaches to model building, often referred to as Group Model Building (Vennix 1996, Luna-Reyes and Andersen 2003, Richardson 1991, all share the same underlying viewpoint that involving client groups and subject matter experts in the model building process, especially in matters of policy and strategy (Luna-Reyes et al 2006) is important but also challenging. One such process that can been used to capture qualitative data is the Delphi method.…”