Overconfidence and Optimism: An Investigation of their Relative Contributions to Bias in Uncertainty Assessments
Daniel Gray, Long Nguyen, Joseph Sarti - 2014
Australian School of Petroleum
University of Adelaide
Cognitive biases are unconscious errors that severely affect the quality of decisions both within the Oil & Gas industry and everyday life. Due to the heterogeneous nature of reservoirs and the limited sampling of their properties, good decision making relies heavily on impartial uncertainty assessments. The biases in these assessments can be reduced to overconfidence: underestimating the true range of uncertainty in our estimates and optimism: deciding on a most likely estimate that is better than what the available data suggests.
This study aims to investigate the relative contributions of overconfidence and optimism to bias as well as their interplay with each other and with other factors. There is an evident absence in the literature on the impact of optimism and previous studies have only investigated overconfidence using fact-based questions where a correct value exists but may or may not be known to the participant. These shortcomings have been addressed via the creation of a survey that elicits comparable conditions to those faced in industry. Participants were required to perform uncertainty assessments for the outcomes of sporting events as well as provide motivational and demographic information.
On average participants produced very poor uncertainty assessments. Relationships were revealed between overconfidence and optimism and with other factors that indicate investment and frequent experience do affect levels of optimism and overconfidence, which has implications for the O&G industry. Developing a better understanding of what influences optimism and overconfidence is vital in eliminating bias from uncertainty assessments and improving the quality of decision making, which in turn has the potential to save billions of dollars on the numerous sub-hurdle rate and sub-economic projects that only proceed due to bias.