Interpreting the results of Kruskal-Wallis QUESTION

#1
If I have used Kruskal-Wallis to compare 3 data sets of diameter values for 3 different populations of trees, and the results are significant (H=19.21, df=2, n=156, p<.0001)
erpreting results
I report that "The mean ranks for diameter are significantly different (H=19.21, df=2, n=156, p<.0001)."

and CAN I ALSO CONLCUDE that: "Average tree diameter is significantly different among the 3 populations." ?

Thanks
 

Dragan

Super Moderator
#2
If I have used Kruskal-Wallis to compare 3 data sets of diameter values for 3 different populations of trees, and the results are significant (H=19.21, df=2, n=156, p<.0001)
erpreting results
I report that "The mean ranks for diameter are significantly different (H=19.21, df=2, n=156, p<.0001)."

and CAN I ALSO CONLCUDE that: "Average tree diameter is significantly different among the 3 populations." ?

Thanks
You're actually testing more than one hypothesis when conducting the KW test. Setting that issue aside, a more accurate way to say what you wrote would be: "The average RANK tree diameter is...bla bla bla..."
 
#3
and CAN I ALSO CONLCUDE that: "Average tree diameter is significantly different among the 3 populations." ?
No, you can't necessarily draw this conclusion. The KW test, as the non-parametric equivalent to ANOVA, also yields an omnibus statistic and a significant result tells you only that at least one of your groups differs from one of your other groups. Like ANOVA, to determine which groups differ requires post hoc analyses.
 
Last edited: