Measures of risk associated to regulations compliance: a laboratory experiment on the use of common-pool resources

The present research examined the influence of different risk perspectives by the use of four different target persons (who could be affected: abstract person, self, specific person, and specific others) and of four different questionnaire answer formats (rating, open percentage, open, and closed frequency) on risk assessments. It was assumed that subjects use two different systems in terms of probabilistic reasoning: a distributional approach for abstract targets leading to higher risk estimates and a singular approach for specific targets leading to lower risk assessments. According to unrealistic optimism (UO) research (showing higher risk assessments for an abstract person than for self), the assumption was that risk assessments for a specific (named) target lead to lower risk assessments compared to an abstract target. Further, common quantitative answer formats for assessing risk were compared to explore differences in risk estimates. The hypotheses were tested using data of a sample of 512 students from a Bavarian university. The frequently confirmed finding of UO could be replicated with both rating scale and open frequency scale, which appeared to be more sensitive compared to the other used scales. However, UO disappeared when the comparison target was specific. Further, risk assessments for an abstract target were highest within every answer format and lowest for specific targets. Furthermore, results revealed that the type of answer format has a moderating effect on the extent of the influence of risk perspective on risk assessments. Overall, this study gives evidence that both the chosen scale and the risk perspective strongly influence risk assessments. Results aim to contribute to the research fields of quantitative assessment of perceived risk. They suggest that probabilistic reasoning in regard to risk not only underlies motivational or cognitive ego-defensive mechanisms but is rather presumably caused by the use of different systems of inferential strategies.