An evaluation of the statistical implication of the violation of regression assumptions

Description

Abstract

An evaluation of the statistical implication of the violation of regression assumptions. This study seeks to evaluate the statistical implications of the violation of regression assumptions. The study utilized a socio-economic dataset in regression modeling with a focus on validating the assumption of residual and normality. From the analysis of data in this study, we utilized two set of data, one original and the other, a copy of the original but contaminated. The result of the study has been analyzed with the regression model, with reference to validating the regression assumptions, specifically on test of normality and the residual analysis. The study result found that both models on the different sets of data indicated that they are adequate. Further analysis revealed that the model with original values satisfied the regression assumption of normality and residual and the coefficient estimate were higher compared to the second model where the assumptions were violated. Despite the indication that the model is adequate, the parameter estimates were insignificant and thus, any inference drawn upon the contaminated dataset might be misleading. Thus, the essence of validating the assumptions of regression model is paramount in empirical research as its violation might be catastrophic.