please consider the following example: #start code set.seed(123) level<-rnorm(18, 10,3) group1<-rep(letters[1:3], each=6) summary(aov(level~group1)) group2<-rep(1:3,each=6) str(group2) summary(aov(level~group2)) #same result as for group1 summary(aov(level~factor(group2))) #same result ad for aov anova(lm(level~group2)) #end code what I would like to do is to perform an anova among groups (analysis of variance for three different gruops); consider that groups are completely arbitrary: they are not intended to have any sort of scaling or ordinal meaning; in my example same groups are coded in two alternative ways: group1 as "chr" (factor) and group2 as "num"; so by keeping in mind my purpose (is there any difference in the level among groups?) I would simply consider the result of aov() for group2 (num) as a non sense (with respect to my specific purpuse) is that a correct interpretation? I hope not having misinterpreted the indications of the following thread http://r.789695.n4.nabble.com/Question-about-factor-that-is-numeric-in-aov-td2164393.html thank you for any help best regards max