iSixSigma

DF for 2 sample t-test with unequal variances

Six Sigma – iSixSigma › Forums › Old Forums › General › DF for 2 sample t-test with unequal variances

Viewing 2 posts - 1 through 2 (of 2 total)
  • Author
    Posts
  • November 3, 2006 at 2:18 pm #45145

    Ropp
    ★ 10 Years ★
    Participant

    With a 2-sample t-test assuming equal variances the total degrees of freedom is simply N-2.  If you can’t assume that the variances are equal the df of course will be lower, presumably because it’s eaten up by having to account for the multiple std dev’s.  I understand why this would be the case but would like to know the formula/calculations of how to arrive at the correct df.
    Thanks,
     

    0
    November 3, 2006 at 2:28 pm #146476

    Robert Butler
    ★ 20 Years ★
    Participant

    The equations are from pp.299-303 Brownlee Statistical Theory and Methodology in Science and Engineering – 2nd edition 1965
    We are assuming you have the two means of interest, their respective standard deviations, and the associated number of samples/mean
      data values1;  input xbar1 std1 n1 xbar2 std2 n2;   rat1 = (std1*std1)/n1;    /*computation of fprime the synthetic 
                                            degrees of freedom*/  rat2 = (std2*std2)/n2;  nval1 = (n1 – 1);  nval2 = (n2 – 1);
      fnumerator = (rat1 + rat2)*(rat1 + rat2);
      fdenom1 = rat1*rat1/nval1;  fdenom2 = rat2*rat2/nval2;
      fprime = fnumerator/(fdenom1 + fdenom2);
      meandiff = xbar1 – xbar2;  /*computation of the t statistic*/  tdenom = sqrt(rat1+rat2);
      tvalue = meandiff/tdenom; 

    0
  • Author
    Posts
Viewing 2 posts - 1 through 2 (of 2 total)

The forum ‘General’ is closed to new topics and replies.