A great deal of research has been directed toward developing intervention strategies for improving safety performance in construction. While the construction research community has become more enlightened about the degree to which accidents can be blamed on workers, nonetheless it must be acknowledged that a certain amount of risk-taking by workers is often involved. This behavior seems contrary to self-interest, because while the consequences to the project or the employer have been measured economically, the consequences of an incident to the worker are more direct, immediate, and severe. In spite of this perceived action against interest, incidents still occur. To consider possible connections between employer attitudes regarding production and reward systems put in place by the employer and resulting worker behavior, an agent-based simulation experiment was conducted. In this experiment, different employer attitudes and reward systems could be modeled and expeienced by a population of workers with variable degrees of native production ability and risk-tolerance, while these workers conduct operations on a simulated project site of spatially-variable danger. By using an agent-based approach, local and random interactions and events can occur and lead to emerging measures for the entire system, in much the same way that local interactions lead to a gross metric such as incident rate. The emperiment demonstrates a link between employer attitudes and reward structures and the distribution of risk-tolerance in the worker population. The impact of interactions between workers and the level of dnanger at the site is considered.
Construction safet, agent-based modeling, simulation, risk tolerance.