I have three mean estimates, all from variables assumed to follow a Normal distribution. The following is a mock example.

x1 is estimated at 4, with a standard error of 5 and 6 degrees of freedom x2 is estimated at 6, with a standard error of 4 and 6 degrees of freedom x3 is estimated at 7, with a standard error of 3 and 6 degrees of freedom

What would be the degrees of freedom attached to (x1+x2+x3) ?

  • $\begingroup$ This depends on the dependence between the three xs. $\endgroup$ Apr 3, 2022 at 19:07
  • 3
    $\begingroup$ How are the estimates made and what exactly do you mean by "degrees of freedom" of such an estimate? (Estimates usually don't have anything known as a "degree of freedom," but often their standard errors do--and even then the meaning of d.f. varies with how the estimate is made.) $\endgroup$
    – whuber
    Apr 3, 2022 at 19:19
  • $\begingroup$ To @RichardHardy: They are independent, I should have stated that. $\endgroup$
    – PerrySun
    Apr 3, 2022 at 20:35
  • $\begingroup$ To @whuber: Those are the DFs attached to the standard errors. They have been estimated using maximum likelihood. If a parameter X follows a normal distribution, it's maximum likelihood estimate follows a T distribution with N-1 degrees of freedom due to the uncertainty on the variance estimate. Those are the degrees of freedom I'm referring to. $\endgroup$
    – PerrySun
    Apr 3, 2022 at 20:38
  • $\begingroup$ @PerrySun, please edit the post to include the additional information from the comments. $\endgroup$ Apr 4, 2022 at 5:01


Your Answer

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