18

I read here that $n-p-1$ was the number of degrees of freedom I should use when doing a t-test for the significance of a regression coefficient, but I don't understand why. My understanding was that t-tests generally had $n-1$ degrees of freedom.

amoeba
  • 93,463
  • 28
  • 275
  • 317
user1205901 - Reinstate Monica
  • 11,303
  • 26
  • 77
  • 152

3 Answers3

19

You lose one degree of freedom for each estimated mean parameter. For an ordinary t-test that's 1 (the mean). For regression, each predictor costs you a degree of freedom. The extra one is for the intercept.

More specifically, the degrees of freedom come from the denominator in the t-test, which is based on the residual sum of squares -- there are $n-p-1$ degrees of freedom in the residual sums of squares.

Glen_b
  • 257,508
  • 32
  • 553
  • 939
  • 4
    In fact if you regress some outcome on a constant (only an intercept in the model) you get the mean, p would then be 0, and one would have n-1 degrees of freedom ā€“ Repmat Sep 28 '16 at 14:32
3

This is not true in general. The number of degrees of freedom of t-test depends on a specific model. They're talking about linear regression. So, t-test for an estimator has $n-p-1$ degrees of freedom where $p$ is number of explanatory parameters in the model.

Community
  • 1
d.k.o.
  • 234
  • 1
  • 6
1

Degrees of freedom is the number of independent values or quantities which can be assigned to a statistical distribution.

So in this case its nāˆ’pāˆ’1 because:

n is number of training samples. p is number of predictors. 1 is for intercept.

Venkataramana
  • 73
  • 1
  • 1
  • 5