TABLE 1.

Descriptions of the modeling approaches included in the review of studies using complex systems models to study retail food environments for population health

Model approach	Description
System dynamics models	An aggregate- (versus individual-) level modeling approach that uses specific techniques (e.g., differential equations, state variables, stocks and flows) to capture and understand endogenous sources of complex system behavior. It centers on the principles of feedbacks and accumulation and is well-suited to simulating and capturing a macroscopic view of system behavior in large populations (24–26, 30, 31)
Discrete event simulations	Simulations with individual actors that are passive entities whose behavior is modeled as a sequence of discrete events in a setting over time. In this approach, events are the priority over the individual entities. Models are commonly used to examine resources and system constraints in meeting a target, and health sciences often use these to determine patient flows through clinical care settings (25, 26, 31)
Network analyses	Models with individual entities (e.g., people, organizations) that measure and analyze the relations and/or flows among them. Models can be used to analyze the network structure as well as how the transfer of information, behaviors, or diseases across connections change as relations for each individual entity change (23–25, 30)
Agent-based models	Simulations with individual actors (i.e., agents) that are active entities which make decisions and/or behave based on a set of rules. Individual actors may interact with each other and their environment and can adapt to these interactions producing emergent properties of the system that make them effective at capturing complex social phenomena (22–24, 31)
Microsimulations	Simulations with individual actors that are passive entities without interactions. Experiments often modify the attribute(s) of individual actors to understand the effect the change has on an individual over time. Common method in the economics field and tends to focus on estimating the detailed predictions of a specific policy/intervention on a target outcome as well as determining its cost-effectiveness (22, 23)
Hybrid models	Combined use of 2 or more model approaches in the same simulation. Offers the advantage of balancing the strengths and shortfalls of each approach to improve the effectiveness of a model in capturing aspects of a complex system (26, 31)