Even when encountering extreme multicollinearity with the use of squared and interaction variables,t tests on the significance of individual independent variables is reliable.

Question

Quizplus · Accepted Answer

When extreme multicollinearity is present, t tests on the significance of individual independent variables can produce unreliable and erroneous results. This is because the coefficients for the correlated variables are difficult to estimate accurately and their significance tests may not reflect their true contribution to the model. In such cases, other techniques such as regularization or PCA may be more appropriate.

Even When Encountering Extreme Multicollinearity with the Use of Squared