Recent blogs on our website had the below information. I don’t understand how if p = 2 the degrees of freedom could be n-3?

Hi,

Given a model which includes quadratic term y=b0+b1x+b2x2+e, if we want to use t test to figure out the p-value of the parameters b0, b1, b2, what is the degree of freedom of the t distribution? n-3?

ReplyQuoteEmail Author

Message ReadMark as Unread Message Not FlaggedSet FlagDecember 6, 2015 8:24:28 PM PST1 day ago Lang Wu UBC InstructorManagerRE: degree of freedom in linear regressionCollapse

Total views: 3 (Your views: 1)

Yes, n-3. The degree of freedom is n-p-1, where p is the number of predictors. Here p=2 since you may view x2=x2