A STATISTICAL STUDY ON THE EFFECT OF MULTICOLLINEARITY ON PARAMETERS OF A MULTIPLE REGRESSION

ABSTRACT

Multicollinearity is a fundamental statistical issue that arises when two or more independent variable in a regression model are highly correlated, posing challenges to the interpretation and reliability of statistical analysis, this abstract provides an overview of multicollinearity, its causes, consequences, and potential remedies. Multicollinearity is often an unintended consequence of including multiple variables in a regression model. It can lead to unstable parameter estimates, making it difficult to determine the unique contribution of each variable to the dependent variable. When multicollinearity is present, coefficient estimates can become imprecise, and their signs may change when the model is altered or when new data is introduced. This purpose of this project is to show if multicollinearity has any effect on the parameter estimates and we analyzed 3 datasets with sizes 14,14 and 92 respectively and it showed that in the presence of multicollinearity the parameter estimates changes as seen from the regression of the second data set, we also saw that a model can still be predicted even in the presence of multicollinearity .Therefore a multivariate regression model with collinear predictor can indicate how well the entire group of predictors predict the outcome variables, but individual predictor may not give valid result.