# Network analysis on JASP

Hello everyone!

I am writing to ask about network analysis on JASP. I and my research group have performed a behavioral study collecting data from three groups of 50 participants (150 in total). Our data consists of accuracy and response time values, which are not normally distributed. What type of estimator do you think that is better to use for our analysis? Is it possible to compare the network plots?

Many thanks in advance,

Elisa

## Comments

Hi Elisa!

Is your data continuous or ordinal? If you have continuous non-normal data you could try using the nonparanormal transformation to deal with non-normality (i.e., after selecting the EBICglasso estimator in JASP, you can select "npn" under correlation method). If your data is ordinal you could try using polychoric correlations, I believe this should be done automatically when you use the EBICglasso estimator in JASP with the "auto" correlation method – notably your sample size is quite small, so you may or may not run into problems when using polychoric correlations depending on how many nodes you have.

You can compare the networks using the Network Comparison Test, but you can only compare two groups at the time. This is also not yet implemented in JASP and can only be done in R at the moment.

Hope this helps!

Adela

Hi Adela!

Many thanks for your help, I'll try this method!

Elisa

Hi everyone,

I had performed a systematic review about medications use for a specific type of surgery.

Each article cited the medications names, and for each article there are at least 2 medications that were compared. Thus, I would like to perform a Network metanalysis of these drug combinations.

I had created a spreadsheet with 3 column (author's name, group 1, and group 2), I put the medications categorized as codes. I have tried to do the network analysis on JASP software; however, when I chose the Network option, the data was not processed.

Is there anyone who coud help me?

Best regards,

Fábio.

Hi Fábio,

I think what you are trying to do is something completely different. The network module as currently implemented in JASP aims to estimate markov random fields and does not really perform network analysis from a graph theoretical point of view.

Best,

Adela

Hi Adela,

Thanks so much for the answers!

I have an additional question, is Likert-7-scale data considered as continuous data?

Best, Xin