Particular risk analysis modelling process

The model of fuzzy risk analysis can be combined with the Dempster-Shafer s model in the risk management process, particularly at the stage... 

In our model of a SHE information system in Chapter 1, we identified four different subsystems, i.e. data collection, data analysis or processing, a memory and distribution of information. We will here first focus on the data-collection subsystem. Chapter 12 presents an overview and Chapters 13 and 14 give details about different methods of data collection in accident and near-accident reporting and investigation, workplace inspections and SHE audits. We will then proceed in Chapter 15 to go through some basic principles for the establishment of a memory (database) on accident risks. After that, we will review different methods for data analysis based on data from accidents and near-accidents in particular. In practice, it is not always easy to separate data collection from data analysis or processing, because these activities often interact. 

Workers will respond to risks not on the basis of their absolute disutility, as in an expected utility model, but in relation to more-or-less fixed standards. These standards therefore take on the function of social norms, patterns of judgment and behavior that, to an outside observer, may appear to refiect identical preferences on the part of each individual, but which, at least in this analysis, simply arise spontaneously in the process of rational collective action. In particular, these norms tend to be comparative rather than absolute, and they depend to some extent on perceptions of intent. The normative selection of risks to be regarded as actionable strengthens the bright line aspect of solutions to the assurance game, while the perception by workers that a risk is deliberately imposed on them by management increases the likelihood that it will be resisted. 

Later in the risk assessment process, once the scenarios of particular interest are clear, FMEA is a very powerful tool for focusing on which component is the trigger in the event and how to make it more robust in the system. The same is true of sneak circuit analysis, cause-consequence analysis, or dispersion modeling. 

First, a careful descriptive analysis of workplace discourse can contribute to the development of a generalized expert model in many important and complementary ways Because the interview process can affect a subject s beliefs and values, researchers must be particularly careful when they attempt to elicit their lay audience s beliefs and misconceptions. If interviews change a subject s knowledge, then interview results will produce a skewed conception of what audiences need to know to understand risk. When interviewers employ a conversational style, for example, they can introduce new words or ideas, but they can also overlook critical concepts. If the purpose of the interview is to discover people s mental models so that communicators can construct risk messages to affect a wider audience, the interview process must not create misconceptions or reshape the subject s mental model. When researchers employ a variety of methods, they must insure that the different forms of inquiry (questionnaire or interview, for example) do not affect their results. Morgan et al. (2002) note that designing good questions is not easy. ... 

---