Multiple Linear regression - comparing standardized coefficients

Question

When performing multiple regression, can standardized regression coefficients (beta's) be used to rank predictors in terms of their predictive importance. That is, do beta's tell you which predictors are the "best" in the model.

Answer 1

In OLS, squaring each standardised beta coefficient can inform you of the proportion of variance that each variable explains in the dependent variable. You can then rank order each variable in terms of their respective $R^2$ values, thus determining which variable explains the highest and the lowest proportion of variance in your DV. In social sciences, it is common to assess the importance of predictor variables in this way.