Bioequivalence and Bioavailability Forum

Dear Ravi,

I’m not gifted with , but it’s important not to assume equal variances (i.e., use Satterthwaite’s degrees of freedom). See also this post.

---

Celebrating login #3000 today.

Dear Ravi,

❝ Can any one please provide me the SAS code for performing One way Analysis of Variance with unequal no. of observations per cell.

Proc GLM data=yourdata;
  class your_factor;
  model your_Y_variable = your_factor;
run;quit;

❝ How it will differ in case of equal no. of observations per cell.

Nothing different.

But as Helmut has pointed out this code assumes equal variances within the groups according to your_factor.
Within the classical ANOVA this assumption can't avoided.
Therefore I don't know any variant of the code with Proc GLM to account for unequal variances for more then 2 groups.
In case of two groups only consider Welch t-test (part of the output of Proc ttest dealing with Satterthwaite approximation).