Briefly discuss approaches that can be employed in situations where multicollinearity is detected

What will be an ideal response?

(a) One can remove an independent variable that is highly correlated with one or more other independent variables. While this will reduce the variance of the coefficient estimate, it can introduce a bias in the coefficient estimate if the omitted variable is important in the model.
(b) One might be able to construct a new independent variable that is a function of several highly correlated independent variables.
(c) One might be able to substitute a new independent variable that represents the same influence but is not correlated with other independent variables.
None of these are always complete solutions in removing multicollinearity. But they all require careful thinking and judgment about the objectives of the model and the problem environment that it represents.