Network analysis JASP

MSC

Hello,

This is the first time I have done network analysis and I have some doubts about performing these analyses and interpreting the results using JASP.

JASP provides an Expected Influence (EI) index. Is it the one-step EI index or the second-step EI index? I believe the R package qgraph provides both...

I am in doubt as to whether I should interpret it as an absolute value. For example, if the EI values for the nodes of a network are as follows:

A -0.385

B -1.726

C 0.575

E -0.311

F  1.964

G -0.603

H -0.200

I -0.011

J -0.421

K -0.263

L 1.687

M -0.304

 

Which of these nodes would have the most influence on the network?

In this case, I would say that the node with the most influence in the network would be node F (1.96), but I am not sure if the second one would be node L (1.69) or node B (-1.73). Due to the presence of negative edges in the network, I don't know if it should be interpreted in terms of positive and negative influence on the network…

 And another question, can I know the stability of this index with JASP? I have seen that the program provides the stability graphs (using bootstrap options) for strength, closeness and betweenness, but not for EI.

Thanks!😊

María

SachaEpskamp

Hi Maria,

I am not a 100% sure on the expected influence variant used in JASP, but I think it is the implementation form qgraph, which is just the sum of weights without taking the absolute value. I personally think that should only be used in some cases where you can reasonably assume all edges should be positive, such as symptom networks. I generally not recommend using expected influence in other graphs where variables could be arbitrarily recoded (e.g., personality items such as "I like to go to parties" could arbitrarily be recoded as "I don't like to go to parties", leading to a big impact on EI).

Having said that, the interpretation then is simply: node F has a lot of positive connections and node B has a lot of negative connections.

Best, Sacha

MSC

Thank you very much for your help!

Best,

María