Estimating a limited population mean for discrete linear regression using some robust regression methods

This study concentrates on estimating the population mean within stratified random sampling utilizing discrete regression estimators, juxtaposed against classical estimators relying on mean squared error (MSE) and efficiency criterion (RE). Despite the ordinary least squares (OLS) estimator's efficacy in accounting for error covariance, it exhibits heightened sensitivity towards outliers. To tackle this challenge, robust regression estimators, such as LTS, Tukey M, and Hampel M, were implemented alongside robust covariance and covariance matrices (MCD and MVE). This study was applied to real data from the Central Bureau of Statistics, specifically the results of the hotel survey in Dohuk Governorate for the year 2020, which included a group of hotels with a size of 115 hotels. By describing the data represented in the total wages paid, the number of employees,  the results showed that the (LTS) estimator achieved high efficiency in estimating the regression parameter compared to the rest of the estimators used in the study