^{a}and S. Rainald

^{a}rainald.kasprik@hs-heilbronn.de

When statistical data from real life experience (e.g. test data or operational data) or from simulator runs is not available, probability assessments are based on subjective likelihood assessments (e.g. Cooke & Goossens, 2000; Dezfuli et al., 2009). This concerns machinery safety (e.g. IEC 61508; EN ISO 13849-1 / ISO TR 14121-2; EN 62061), the security of installations (e.g. Bompard et al., 2007) as well as the management of risks in general (e.g. Murray et al., 2011).

In two empirical studies, the probability ranges of likelihood expressions were elicited for a 5-point scale of probability expressions. Previous studies did not examine probability ranges for a 5-point scale - although this scale is broadly used within risk management and especially in EN ISO 62061 for determining the Safety Integrity Level.

In the first study, odds ratios of an event occurring with regard to the total number of similar situations, were used as independent variable. In the second study, the same odds ratios were used as stimuli but then describing naturalistic random events. Both studies revealed robust probability ranges and a good interrater homogeneity. Naturalistic random events lead to a slightly higher assessment of likelihood of occurrence.

The study's results can help experts scrutinizing their subjective likelihood judgments by feed backing to them the corresponding probability expression (study 1) and by feed backing to them naturalistic random events that correspond with a probability expression (study 2).