Premium
Risk matrix input data biases
Author(s) -
Smith Eric D.,
Siefert William T.,
Drain David
Publication year - 2008
Publication title -
systems engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.474
H-Index - 50
eISSN - 1520-6858
pISSN - 1098-1241
DOI - 10.1002/sys.20126
Subject(s) - econometrics , bayesian probability , statistics , matrix (chemical analysis) , mathematics , psychology , computer science , composite material , materials science
Risk matrices used in industry characterize particular risks in terms of the likelihood of occurreRisk matrix input data biaseshe actualized risk. Human cognitive bias research led by Daniel Kahneman and Amos Tversky exposed systematic translations of objective probability and value as judged by human subjects. Applying these translations to the risk matrix allows the formation of statistical hypotheses of risk point placement biases. Industry‐generated risk matrix data reveals evidence of biases in the judgment of likelihood and consequence—principally, likelihood centering, a systematic increase in consequence, and a diagonal bias. Statistical analyses are conducted with linear regression, normal distribution fitting, and Bayesian analysis. Evidence presented could improve risk matrix based risk analysis prevalent in industry. © 2008 Wiley Periodicals, Inc. Syst Eng