On interval estimation in linear relationships with heteroscedastic measurement errors in both axes

This article investigates confidence sets for linear model with heteroscedastic measurement errors on both axes. We propose several approaches to find confidence intervals for the slope and joint confidence regions for both the intercept and the slope. The performance of the confidence set is compared in terms of both coverage probability and its length/area via theory and simulation studies. Application of these methods is also illustrated with real data sets. Guidelines for choosing suitable confidence set are provided.