Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Questions about Network Analysis in JASP

Hello!

I'm doing my very first network analysis, which I've been wanting to learn for ages. I'm very excited.

I have run the EBICglasso and also a correlation network on my data. In both cases there is at least one single node which is not connected to any of the other nodes. My closeness is zero for both groups which I am comparing. Is the reason that my closeness is zero that there is at least one unconnected node?

Kind regards,

Anna

Comments

  • Hi Anna,

    The closeness is defined as 1 / <sum of all shortest path lengths from one node to all other nodes> for each node. If you have a disconnected graph, then the path length from the disconnected node to the other nodes is infinite, and 1 / infinite -> zero. Hence all closeness measures will be zero then.

    Best, Sacha

    Thanked by 1eniseg2
  • Hi Sacha,

    thank you very much! Now I have another question. I've produced the bootstrapped Centrality Stability plots with case dropping. My healthy control group plot shows only what happens when the sample is reduced from 100% to 85% but with no further reduction. My patient group's plot goes from 100% to 30%. Why are the plots so different, and how can I compare them now?

    Best,

    Anna

Sign In or Register to comment.