I have a linear model and when i run the stepwise variable selection, i got:
Start: AIC=166.27
Y ~ A + B + C + D + E
Df Sum of Sq RSS AIC
<none> 1466.0 166.27
- B 1 81.08 1547.0 166.64
- D 1 436.35 1902.3 175.73
- A 1 498.93 1964.9 177.16
- E 1 559.51 2025.5 178.49
- C 1 1334.19 2800.2 192.74
but its summary states that:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 65.41682 9.93755 6.583 9.07e-08 ***
A -0.22676 0.06305 -3.596 0.000916 ***
B -0.33141 0.22860 -1.450 0.155333
C -0.94235 0.16024 -5.881 8.30e-07 ***
D 0.10382 0.03087 3.363 0.001769 **
E 1.38058 0.36252 3.808 0.000496 ***
B is actually not significant!
