Abstract
The paper focuses on definitions of risk. Six studies were analyzed with respect to how non-expert subjects defined the concept of risk. In five of the studies subjects also rated risks of various adverse events with respect to personal risk, risk for people generally, and threat to society. Data from an additional study allowed a comparison between responses to a unidimensional scale measuring subjects' risk definitions and four suggested risk definitions. The results showed, firstly, that subjects rated risk significantly different when utilizing different perspectives. Thus personal risk was rated lower than risk to people in general, which in turn was rated lower than risk regarding the threat to society. Secondly, on the basis of two subsequent categorizations of the responses to the suggested risk definitions, the results showed that naive subjects did not usually employ a logically coherent definition of risk which exclusively involved the probability and consequence elements. Thirdly, subjects who could be accommodated within logical risk definition categories differed in their ratings on the basis of definition preference. Subjects who focused on a probability definition of risk gave lower risk ratings than those who used a risk definition which focused on consequences of an event. Fourthly, the comparison of a unidimensional rating scale and four separate definitions of risk indicated that the use of a unidimensional scale for measuring subjects' risk definitions may conceal a great variation of responses, especially with respect to the middle response alternative. The results are related to some diverging examples of how researchers define risk, and discussed in a framework where the different ratings emanating from the use of different perspectives are related to concern and behavior. It is suggested that differences between novices' and experts' risk assessments may, in part, be related to clarity, as well as type, of personal risk definition. Risk perception research concerned with low probability/high consequences events is finally related to research which also involves high probability/low consequences events, e.g. in marketing, and it is suggested that future risk perception research might benefit from considerations of both approaches.