The use of the (Tobit) Model in Studying the Variables affecting the Increase in the Number of People with Systolic Hypertension, in the Presence of the problem of Heterogeneity of Random Error Variance

Abstract

The process of regression analysis is based on a number of basic assumptions, and if one of these assumptions is not available, this will lead to obtaining inaccurate results, and the most of these assumptions is homoscedasticity, violating this assumption causes a problem of heteroscedasticity of the random error variance the, which may come from different variances This cause to misleading and inaccurate decisions.

      Recently Econometric models received attention, especially limited regression models, which contain specific response variables and several repeated observations within a certain range, including Tobit models and also considered imitations of the Censored Regression Model. Problem arise when it appears in the data heteroscedasticity, which makes methods of estimating the parameters of the linear regression model giving incorrect and inaccurate results, hence we get Biased parameters that do not have minimalist variance, as well as getting an unreliable p-value.

      The objective of this research is to study the estimation of the multiple standard (TOBIT) model which features that the variable (Y) specific variant. In heteroscedasticity, which both likelihood function method weighted and Genetic Algorithm   was used in simulation.

       To sum up, the results showed, that Genetic algorithm better outcomes than weighted likelihood function method therefore, it was used in the practical application.