An errors-in-variables model based on the Birnbaum–Saunders distribution and its diagnostics with an application to earthquake data

Regression modelling where explanatory variables are measured with error is a common problem in applied sciences. However, if inappropriate analysis methods are applied, then unreliable conclusions can be made. This work deals with estimation and diagnostic analytics in regression modelling based on the Birnbaum–Saunders distribution using additive measurement errors. The maximum pseudo-likelihood and regression calibration methods are used for parameter estimation. We also carry out a residual analysis and apply global and local diagnostic techniques in order to detect anomalous and potentially influential observations. Simulations are conducted to validate the proposed approach and to evaluate performance. A real-world data set, related to earthquakes, is used to illustrate the new approach.