Bayesian Analysis of Strength Properties of Particleboard

Abstract

Strength properties (modulus of rupture, internal bond strength, lateral nail resistance, and nail-head pull-through) of particleboard were measured, and were subjected to conventionally statistical and Bayesian analyses. All strength properties could be fitted to normal distributions. In conventionally statistical analysis, mean and variance are constant values that have one true value. Therefore, when mean and variance are distributed, probability distributions of strength properties could not be inferred suitably using conventionally statistical analysis. In Bayesian analysis, because mean and variance are considered to be distributed, their distributions could be used to infer probability distributions of strength properties. The mean and variance of products with large variance of strength properties, such as particleboards, are considered to be distributed. Thus, Bayesian analysis is useful in the quality control of particleboards.