Difference between "rand NW" and "permutation" in GLM panel

Submitted by veraP_20 on

Dear GraphVar team,

I want to use a nonparametric test to do a group comparison, but I don't know which of the two options of non-parametric test would be apropriate. Could you explain me the difference between them? I think in my case it would be the "permutation" option , but I would like to be sure.

All the best,


Johann Kruschwitz

Thu, 08/03/2017 - 11:11

Hi Vera,

the two options refer to non-parametric testing on the network level (random networks) and non-parametric testing on the metric level (permutation). For a non-parametric group comparison I would chose the regular permutation testing as otherwise you would compare the graph metrics  - derived from random networks from within each group - between the two groups and determine were the "real" group difference would be situated in this null-distribution. There could be hypotheses for that but this would be a rare case. Random networks is a good choice when performing e.g. correlational analyses.