Doctoral Dissertations

Date of Award

12-2024

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Industrial Engineering

Major Professor

Mingzhou Jin

Committee Members

James Ostrowski, Andrew J. Yu, Shuai Li

Abstract

The topic of motivational bias in expert risk estimation for dams in the United States under the control of several federal agencies is addressed in the guidance documents of those agencies. However, the description of motivational bias in those guidance documents is limited to those situations wherein an expert risk estimator has a vested interest in the outcome of the risk assessment, such as some financial incentive to perform remedial work at the facility being assessed. This description is consistent with the literature upon which the guidance documents are based. Subsequent literature provides alternate descriptions of motivational effects that may also influence the behavior of the expert estimators. To determine if this is the case, a survey was developed to answer research questions related to this subsequent work and was distributed to dam safety industry professionals. This work provides evidence that these effects are likely influencing the dam safety estimation process and suggests training and facilitation techniques to reduce or eliminate their impact.

No prescriptive process for aggregation of expert risk estimates is provided in the federal dam safety guidance documents. This work presents commonly used aggregation techniques and alternate techniques as described in the academic literature. Decision makers working in the dam safety industry were consulted to better understand the current state of practice and the decision makers' desires with regards to expert weighting, information loss in aggregation, and other practical considerations. Additionally, a measure for determining how well each expert is represented in the aggregated probability function using Kullback-Leibler divergence is presented along with an optimization method to adjust expert weights to improve representativeness in the aggregated distribution. Finally, a method for determining the relative importance of outlying estimates based on the conservativeness of individual expert estimates versus the general tendency of each expert is provided.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS