Hi, usually the first level of the factor is used as the reference ("control") group.

You can specify the order of the levels explicitely by using the argument "levels" when creating the factor with factor(.). When you only want to ste the reference level without changing (or caring about) the order of the remaining levels, you can use the function relevel(.)

Examples:

grp_string = c("control", control", ..., "A", "A", ..., "B", "B", ...)

grp1 = factor(grp_string)

# levels: A, B, control ## note: alphabethic order of the levels

grp2 = factor(grp_string, levels=c("control","A","B"))

# levels: control, A, B ## note: order of levels as specified

grp3 = relevel(grp1, "control")

# levels: control, A, B ## the level "control" is now placed at the first position

Thank you very much to both!

I'll try!

We also saw there is an option to do a similar Many-To-One comparisons by the Xin Gao's non-parametric multiple test: 

http://finzi.psych.upenn.edu/library/mutoss/html/gao.html

Dear Lur,

concerning multiple comparison procedures in R I would suggest the multcomp package. There is a nice book "Multiple Comparisons Using R" about this package.

I also attached an example form the book showing how to perform a Dunnett test and also how to specify (change) the control.

Best, Stephan

https://cran.r-project.org/web/packages/multcomp/index.html

Thank you very much, looks great! I should find for non parametrics too.