Do the partial F-statistic and t-statistics have to agree? For example, consider adding two explanatory variables W and U to a regression. In words, explain why it is possible for the partial F-test to reject H0: bW = bU = 0 but for neither t-statistic to be statistically significant (and hence not reject either H0: bW = 0 or H0: bU = 0). Exercise 41 has a numerical example.