This can be problematic because it undermines the statistical significance of an independent variable. 1While it may not necessarily have a large impact on the model's accuracy, it affects the variance of the prediction and reduces the quality of the interpretation of the independent variables. 1