Hi there,
I am a biomedical engineer and recently sent a manuscript to a medical journal for review. I evaluated the effect of two potential drug compounds on the blood pressure in an animal model.
My research is as following:
Group A: control animals (no treatment at all).
Group B: animals receiving compound A.
Group C: animals receiving compound B.
I did ANOVA and three Tukey post hoc analyses to compare all the groups, using SPSS. The SPSS output showed that the group C is significantly lower than A and B.
The stat reviewer pointed out that, for the Tukey test the p criteria (0.05) should be adjusted downward to 0.05/3, since I made three comparisons.
However, didn't the textbook on Stats say that the Tukey test already took the nature of multi-comparisons into consideration so that the overall family-wise alpha is ensured to make a overall p of 0.05? Or it is simply like a t test for multiple comparisions, which requires Bonferroni correction? Or, at least, SPSS does not do that adjustment automatically?
After googling for minutes, I found websites that suggest contradicting informations.
I am pretty sure a lot of experts are out there who knows this well. Could you give me a hint?
Thanks very much.
R Dawson
Univ of Victoria, Canada
---------------------------------
All new Yahoo! Mail
---------------------------------
Get news delivered. Enjoy RSS feeds right on your Mail page.
|