PhD Seminar in Quantitative Marketing Research

System dynamics (SD) is a well-established approach within the systems sciences for the purpose of modeling and simulating dynamic (i.e., changing) and complex (i.e., caused by many interrelated variables) policy problems (Sterman, 2000). The goal of SD is to design new policies that are effective in the short and long-term to control or “solve” the problem of concern. Key applications of SD can be found in sustainability research (circular economy) and health care (integrated health care). The policy design process, however, is challenging particularly in comprehensive SD models with many model variables and relationships. Recent research demonstrated that measures from network controllability can help in identifying influential “nodes” in SD models that can be the foundation for policy design (Schoenenberger et al., 2021). Among network controllability analytical tools, the focus so far has been on the control centrality measure (Liu, Slotine, Barabasi, 2012) to be applied to a graph/network representation of SD models. In this context, Schoenenberger et al. (2021) developed a software (called “Executive System Analysis”) to integrate network analytical tools into the SD modeling and analysis process. The software is implemented in Java using the Eclipse Rich Client Platform (RCP). The goal of this thesis is to develop further the Executive System Analysis software in order to systematically evaluate the usefulness of network analytical tools for the analysis of SD models. More specifically, this thesis entails the following steps:  (i)	“Quick” literature review on measures/indices in the area of network control, and similar fields in network science.  (ii)	Identification of “promising” network analytical tools for the analysis of SD models. Implementation of such tools into the existing software application “Executive System Analysis”. Priority should be given to the integration of the package “CalControlCentrality” into the software. The “CalControlCentrality” package was developed by Liu, Slotine, and Barabasi (2012) and is currently hosted by Harvard University. “CalControlCentrality” calculates control centrality indices for all nodes in a given network/SD model.  (iii)	Testing of the implemented network analytical tools on a sample of well-known SD models (iv)	Evaluating of the potential of network analytical tools for the analysis of SD models  Required skills: programming (ideally first experiences in Java programming), interest in network science/network controllability and systems sciences  References: Liu Y-Y, Slotine J-J, Barabási A-L (2011). Controllability of complex networks. Nature, 473:167–173 Liu Y-Y, Slotine J-J, Barabási A-L (2012). Control centrality and hierarchical structure in complex networks. PLoSONE, 7:e44459 Schoenenberger L, Schmid A, Tanase R, Beck M, Schwaninger M (2021). Structural analysis of system dynamics models. Simulation Modelling Practice and Theory, In Press. Sterman J. Business dynamics. Systems thinking and modeling for a complex world. Maidenhead, Berkshire: McGraw Hill Education; 2000

During the next PhD Seminar, 18th May, Claudia and Anne will be presenting their project. The title of the presentation is "Sharing Data for Social Good: From Uninformed Consent to Misinformed Dissent".

When making the decision to use a service for personal benefits, consumers are fast to underestimate privacy-related costs and hence, freely share their personal data (uninformed consent). This cost-benefit analysis shifts, when focusing on data sharing for a social good. We show that asking people to use a service that serves a social good (containing the spread of the coronavirus), they overestimate the costs and rather not use the service due to their privacy concerns (misinformed dissent). To increase data sharing for a societal cause, we test two interventions on how privacy-related information should be communicated. Our results indicate that additional privacy-related information on a service (1) is not processed thoroughly when consumers already have a strong prior conviction about using the service; (2) increases knowledge and positive attitude only if the information is processed thoroughly; or (3) information is presented in a comparative manner compared to single information. In addition to the privacy-related cost perspective, we further show that consumers need be presented with either a (1) direct or (2) immediate personal benefit to share their personal data for a social gain.


Jasmin De Clercq