Help me for a Fisher test : i want to compare 3 groups and proove there is no significant difference

[Cyril Andrieux]

Delete this post please... I found.

many thanks.

[Marcos Almeida]

Welcome to the Stata Forum.

I'm not 100% sure, but I guess the post #1 cannot be deleted.

After the one-way ANOVA, to perform the post-hoc Fisher test (I assume you mean the post-hoc Fisher-Hayter) you will probably need the user-written fhcomp (by Philip Ender).

Just as a side note: you said you "wish to prove there is not significant difference", but this seems, well, odd, to say the least, because we do not "prove" the null hypothesis under a frequentist frame. What is more, this model may be facing "too much noise", for the SDs are quite large. Maybe there is skewness as well, who knows. Moreover, we see an amazingly small R2, which doesn't favor the potential claim that the "adjusted" model would be "explanatory", so to speak. Furthermore, the residuals represent around 90% of the partial SS. Last but not least, a negative value for the adjusted R-squared indicates the model might eventually be considered "technically wrong".

Hope that helps.

[Nick Cox]

You can't delete posts or threads. Asking to do that misunderstands the point of a forum.

A forum like this exists, mostly, to post questions that might be of use or interest to others. That's why questions are asked and answered so that others can see them. (There can be other good reasons, e.g. to announce new programs, and the same applies there too.)

Treating it as a means of getting personal answers in public, which don't matter once you have them, misses the point.

[Carlo Lazzaro]

Cyril:
as an aside to previous helpful comments, posting the solution you've found would be highly welcomed, as others on the forum, challenging with the same issue, might benefit from it.