Can we use the l.s.d (least significant difference) or SED (standard error of difference) for several treatments to calculate the average SE (standard error) for these treatments?
Why do you want to calculate average SE. To me, it does not serve any purpose. If you are interested in pairwise comparisons after significant ANOVA, there are many post-hoc tests available which work under specific assumptions/situations related to data. Select the most appropriate one.
Thanks Dr. Murali Dhar. I want to know that can I calculate the average SE based on the l.s.d or sed from published studies. After checking for several days, I think till now I can not do this.
Now, you have clarified the issue. yes you can estimate SE based on cd value. Cd is being calculated as
CD = (Sqrt (MSE/r))* (t-value at 5% and df).
The only issue is to know the value of r. The df is error degree of freedom and 5% is level of significance. For more than 30 df, at 5% the t-value if 1.96.
You can know r, simly by counting the number of observation, based on which each means (which is under comparison) is estimated.