2019 Volume 65 Issue 2 Pages 83-92
Medium density fiberboard (MDF) was subjected to outdoor exposure for 21 years to evaluate its durability, and reduction in the modulus of rupture (MOR) was then inferred using Bayesian analysis. The durability of MDF after outdoor exposure of 21 years was evident. Different posterior predictive distributions of MOR were apparent for each exposure time due to irregular variation of variances in MOR. As a result, it was not possible to adequately infer the posterior predictive distributions of MOR using Bayesian analysis with a model that included constant MOR variance. The irregular variation of variances in MOR was therefore considered to be due to random effects, and a hierarchical model that included these random effects was constructed: the posterior predictive distributions of MOR were then adequately inferred using Bayesian analysis with the hierarchical model that included the random effects. In addition, 95% prediction intervals were inferred using the model including constant variance in MOR or the hierarchical model including the random effects; however, the 95% prediction interval of the latter model was found to be more suitable than that of the former model.
