Learning for safety: Improvements of Swedish authorities’ toolkits for societal resilience

Abstract: This thesis suggests improvements of selected parts of the Swedish authorities’ toolkits for societal safety and crisis management; crisis response evaluations, crisis management exercises and organizational risk assessments. The thesis also explores how visualizations of safety culture data can be used to support safety culture development. The research was motivated by practical needs and delivers results that can be used to facilitate and improve efforts for societal safety and crisis management. Empirical data has been collected from five Swedish public organizations (three municipalities and two county councils) through interviews, observations and questionnaires. Most of the research has been performed in close cooperation with practitioners. Methods from design science have been used to arrive at applicable solutions to the practical problems motivating the research. The thesis shows how learning results with broader applicability can be achieved from the evaluations of singular crisis responses. Evaluations of crisis responses do not necessarily have to focus on as accurate accounts as possible of what happened. To support the development of crisis management capability they should instead revolve around alternative possibilities. From a summary of what actually happened the exploration of possible variation can bring about broadly transferrable learning results. Evaluation results and explorations of variation should be disseminated throughout the organization. Crisis management exercises often produce vague results with unnecessarily limited applicability. This thesis presents a framework that can help to strengthen the learning effects of discussion-based crisis management exercises. In preparing exercises, aspects of reality that are considered relevant in future instances of crisis management should be identified. Some of them should then be used as parameters in a scenario description. In discussions, exercise participants should collectively alter the parameter representation of the scenario. This can establish shared mental models and provides variation for the individual participants to experience. Experiencing variation is vital for learning and developing capability. Important principles for the design of organizational risk assessment systems for large organizations are forwarded. Large organizations are typically hierarchically layered and laterally split into thematic areas. With such structures, first order analyses pertaining to single organizational units and their areas of operations should be performed unit-wise on all levels, and second order analyses with a systems perspective should be performed for all aggregated subsystems up through the composite organization. In second order analysis, data from the first order analyses of constituent organizational units needs to be reanalyzed, with level-appropriate questions and methods. It is not sufficient in a second order analysis to simply add or aggregate information from the first order analyses of the units in the system, and additional input may also be required. Organizational risk assessment in large organizations faces many communicational challenges, which pose major threats to the functionality of the risk assessment systems. This thesis presents countermeasures to such communicational challenges. For example, efforts to create and use shared knowledge, the bridging of steps of formal communication, the use of dialogue, and the standardization of parts of communicational work can help to reduce the threat of miscommunication. An organization’s safety culture can be developed through emergent change, which requires that relevant information is available to the organizational members. To support such change processes presentations of collected safety culture data should preferably: Facilitate the comprehension of data; Offer suitable relevance structures to the target group; Provide possibilities to experience variation; Evoke inquiry and inspire hypothesizing; and Visualize relations between different parts of data.