Coherent and Compatible Statistical Models in Structural Analysis

Modeling problems in structural analysis requires of a statistical approach that allows the consideration of the random nature of the variables as well as the uncertainties involved in the problem analyzed. However, neither all statistical models are valid nor all assumptions are mathematically or physically reasonable. The aim of this paper is twofold: (a) to explain how to build statistical models with mathematical and physical coherence, and (b) to describe the most common mistakes made when building or selecting mathematical and statistical models. Some interesting tools are provided to carry out this important task and some examples are presented showing the inconveniences and consequences derived from an incorrectly established model.