1
$\begingroup$

I know that the standard error is calculated from looking at the sum of squared errors divided by square root of n.

However, as we add more and more variables, we expect the sum of squared errors to decrease and eventually overfit. So I expect to see the standard error to decrease with each variable on the anova table but it doesn't. enter image description here

Is it fitting a single variable each row or is the error calculated from the prediction with all previous rows too?

$\endgroup$
1
  • $\begingroup$ If you add more variables to a model you would in general expect the standard errors of the coefficients to increase so I am not sure what you are asking. $\endgroup$ – mdewey Sep 29 '19 at 14:28
0
$\begingroup$

That looks like a table of jointly estimated coefficients, not an anova table.

This means the standard errors are all about the same value. You can also interpret the wald statistic and p-value as being "marginal" tests/effects. That is, if that predictor was last in a "sequential" anova table.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.