# Does anyone have any rules for degrees of freedom?

I just did the quant eoc #9 on page 323 and I figured I would use 11 degrees of freedom since there are 12 observations and one independent variable. The answer says to use 10 degrees of freedom without explanation. I’m assuming its because I was supposed to think n-k-1. When do I use n-k-1, n-1 and n-2?

DW and F-Test use k degrees of freedom in numerator, F-test uses n-(k+1) in denominator and DW uses n. T-stat uses n-(k+1) and Chi squared uses k. Can’t think of any others at the moment.

N:B k is no. of independent variables

The number of degrees of freedom is the number of independent variables _ data points _ you have less the number of parameters you estimate.

If you have 12 data points and you create a regression line with two parameters (slope and intercept), you have 12 – 2 = 10 degrees of freedom.

I think you meant to say the number of observations (cases) less the number of estimated parameters gives the degrees of freedom.

I did, indeed. (Note that I got it correct in the example.)

I’ve fixed the errant noun phrase. Thanks for catching that.

Yes, your example was spot on. I figured that you just made a typo, but that I should point it out in case there was an unsuspecting reader!

We all make mistakes. I want people to catch my mistakes so that I can correct them. As you say, we don’t want the candidates being misinformed.

Thanks again.