Fitting Regression Models When Both Variables Are Subject to Non-normal Error Terms

This paper explores the structural measurement error model under the assumption of non-normal distribution for the error terms, specifically following the Exponential distribution. The article considers the use of non-identical distributions for the error terms and employs the Method of Moments (MOM) to estimate the unknown parameters based on these pre-assumptions. Real data on the relationships between Man-Hours and Workload are utilized to evaluate the performance of the estimators. Additionally, the precision of the estimators is compared with Wald-type grouping estimators. The findings of the analysis reveal that the grouping methods outperform the MOM in terms of fitting the non-normal structural measurement error model, as evidenced by lower mean squared error (MSE).