Breaking the Same Board Twice: The Magic of Statistics!

• Activity Number: 566
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Risk Analysis
• Author(s): Yanling Cai*+ and James V. Zidek
• Companies: University of British Columbia and University of British Columbia
• Keywords: size censored measurements ; penalized maximum likelihood ; power prior ; lumber strength ; reliability ; competing risks

Abstract

Stresses affect strength properties of a lumber specimen simultaneously. A strong stochastic dependence between these properties would diminish a wooden structure's reliability. Assessing strength relationships is a seemingly insurmountable challenge as every one of these strengths needs a destructive testing to measure. Here is where statistics comes to the rescue. The paper describes a large-scale experiment designed and conducted by Statistics graduate students in a commercial testing lab. It proposes the new theory to infer the stochastic dependence, which applies proof load technique with a penalizing likelihood approach motivated by a Bayesian conjugate prior. The result: asymptotically consistent, more efficient.