Determining t-values in multiple regression without STDERR(parameter)

86 Views Asked by At

If one of the t-values (and its SE Coef) was erased from the output below, how could you still determine its value, from other output shown?

Output:

The regression equation is Cons = 29.6 + 0.95 Price + 0.777 Temp - 0.0691 TempPrice

Predictor      Coef  SE Coef      T      P 
Constant      29.59    51.03   0.58  0.568 
Price         0.955    5.648   0.17  0.867 
Temp         0.7773   0.8778   0.89  0.386 
TempPrice  -0.06908  0.09707  -0.71  0.485 

S = 7.50216 R-Sq = 24.9% R-Sq(adj) = 13.6%

Analysis of Variance

Source          DF       SS      MS     F      P 
Regression       3   372.98  124.33  2.21  0.119 
Residual Error  20  1125.65   56.28 
Total           23  1498.63

Source     DF  Seq SS 
Price       1  198.38 
Temp        1  146.09 
TempPrice   1   28.51

I know $t=\frac{parameterEstimate}{SE(parameter)}$ ; however, since $SE(parameter)$ is removed, I'm not sure how to recalculate it.

In simple linear regression, $t=\sqrt{F}$ ; however this regression model has 3 predictors, so I don't think that relationship holds anymore.

I also thought to find the t-stat from a t-table, since I know the p-value. However, here the p-values are all very high, and the degrees of freedom for the test is 20, so the sample t-table doesn't go that high. Besides, that seems more of a hack than an actual answer.

Also, I know that in simple linear regression, $stderr(b_1)=\frac{s}{\sqrt{Sxx}}$ ; but, again, I don't think this is applicable to multiple regression.