Generate null-model networks: allows creating subject specific null-model networks
(with the same settings as the original networks – i.e. threshold range, number of nodes) that can
be used for normalization of the network topological measures (including calculation of small-worldness), and for non-parametric testing of
the significance of correlations among network measures with user defined variables (see “Test against random networks (Correlation)”).
The number of subject specific random networks to be created and the
function used to generate such a network depends on your
input (randomizer_bin_und, randmio_und, randmio_und_connected, null_model_und_signed;
please refer to the description of these functions on the website of the brain connectivity toolbox; please also refer to“Appendix 2:” for how to
include your own functions for null-model network
generation into GraphVar). The number of iterations refers to the second argument in the
randomization functions (e.g., to “swap_bins” for the “null_model_und_sign” function).
For reliable and valid assumptions using null-model networks it is advisable to generate a high
number of reference networks. Reference networks created with this option are not used as
reference association-matrices as needed for computing network based statistics
(see “The Raw Matrix (link wise) panel for info on network based statistics).