Is Fríedman one-way or two-way ANOVA test by ranks?

[Dolphin72]

Hi,

Friedman test is frequently called a one-way or two-way analysis of repeared measure ANOVA on ranks. But what is it, similar to one-way or two-way? Can I get information on interactions?

I am also keen to learn the differences between the Friedman and Quade test? The latter seems to be an extension of the Wilcoxon signed rank test invented by Dana Quade in 1979.

Many thanks

Dolphin72

[Dragan]

Well, the dirty little secret is that the basic nonparametric procedures are (basically) really nothing more than applying the usual parametric (t or F test) to the ranks of the data.

That said, you're going to run into trouble with nonparametric procedures that involve interactions. Remember, ranking is a non-linear transfromation. When you rank the data, you can artificially create interactions, when there were not any in the original data points..e.g. factorial ANOVA.

[Dolphin72]

Hi Dragan

many thanks for your reply. So do you think calling the Friedman test as an equivalent to two-way repeated measure ANOVA is misleading? I find it quite often in the literature, but describing it as one-way repeated measure ANOVA is more frequent.

Dolphin72