adjusted p-value = p-value*(total number of hypotheses tested)/(rank of the p-value)

I was taught that "corrected p-values" are not fully justified, because you loose their primary meaning. Instead I use just "corrected significance level". I calculate everything with the spreadsheet by Manuel Weinkauf:

https://sites.google.com/site/weinkaufmanuel/programs/excel

As with Michal I would adjust the threshold for significance

Hi Joe.  See John McDonald's explanation (with example) at the link below--scroll down to Controlling the false discovery rate.  HTH.

http://www.biostathandbook.com/multiplecomparisons.html

Many thanks all for your valuable input!

You can also use the simple R procedure for adjusted p-values

p.adjust(c(.005,.009,.05,.1,.2,.3),method="hochberg")

Thanks Adrian - will give that a go!

The adjusted p-values for the Hocjberg procedure are (qi) defined recursively beginning with the largest. For m comparisons, qi=p(m) for i=m and qi=min{qi+1, (m-i+1) p(i+1)} for i=m-1,...,1 where p(1)

adjusted p-value = p-value*(total number of hypotheses tested)/(rank of the p-value)

HI All - apologies for slow responses on this, but thanks for all your advice!

I have a question.

Why there are articles with tables collecting p values calculated by Tukey test and some of them by Benjamini? Which is the correct procedure? First Leven's test, second Tukey of the significant p values and Third Benjamini of all values (significant or not by Tukey? ), then some of the p values can be not significant by Tukey but they can be with Benjamini? Please, help me :-)