For a regression model of executive compensation, you are given:
| Executive Compensation | ||||
| Coefficients: | Estimate | Std. Error | -statistic | -value |
| (INTERCEPT) | –28,595.5 | 220.5 | –129.7 | <0.001 |
| AGEMINUS35 | 7,366.3 | 12.5 | 588.1 | <0.001 |
| TOPSCHOOL | 50.0 | 119.7 | 0.4 | 0.676 |
| LARGECITY | 147.9 | 119.7 | 1.2 | 0.217 |
| MBA | 2,490.9 | 119.7 | 20.8 | <0.001 |
| YEARSEXP | 15,286.6 | 7.2 | 2132.8 | <0.001 |
Determine which variable or variables should be removed first prior to rerunning the model.
Copyright 2023. The Society of Actuaries, Schaumburg, Illinois. Reproduced with permission.
Key: C
The variables TOPSCHOOL and LARGECITY both lack significance and are candidates for removal. However, only one variable should be removed at a time, and it should be the one with the highest p-value, which is TOPSCHOOL. After removing TOPSCHOOL and rerunning the model it is possible that LARGECITY will become significant.
Copyright 2023. The Society of Actuaries, Schaumburg, Illinois. Reproduced with permission.