network analysis
I am working with JASP version 0.19.3. I tried to run a classical network analysis, but I run into several errors:
1. When I want to classify groups with colors in the graphics options, it goes wrong. The variables get the wrong color.
2. When I run bootstraps I get peculiar graphs, which do not seem to be correct
3. The weight matrix doesn't match the shown edge labels in the network.
I am having trouble figuring out what is going wrong. Do you guys have any ideas?
Comments
Hi,
Thanks for posting the issue! Would it be possible to share the jasp file or a reproducible example so that I can try to figure out what goes wrong?
Cheers
Don
Hi Don,
Sorry for not replying to you earlier. I hereby send you a simplified dataset: Can you read it? If not, how am I supposed to share the file?
Again running into errors in the analyses:
1. The item Gangst en KWL get the wrong colours in the network
2. The weight matrix does not always match the numbers in the network (see example Gboos-Gangst)
3. Again, my bootstrap at the bottom is different from what I am used to. I don't understand how to interpret it.
And a last question. Is it also possible to
Is it also possible to calculate from the expected influence the centrality stability?
I hope you can help me.
Cheers,
Marieke
Hi Don,
Another question: a colleague of mine is still working with the older version of JASP. In that, the network analysis does work fine. Is it perhaps possible for me to access that older version? I need it for a research project for my study.
Thanks!
Marieke
Previous versions are available here: https://jasp-stats.org/previous-versions/
Hi Marieke,
Thanks for bringing this up. Since jasp 0.19.1, it appears that the column names are ordered incorrectly in the weights table and also when coloring the nodes according to a group. I've fixed this in https://github.com/jasp-stats/jaspNetwork/pull/141 and https://github.com/jasp-stats/jaspNetwork/pull/145, so this should be fixed in the upcoming version.
In the meantime, jasp 0.19.0 should not have the same bug, so you could use that version.
Best,
Don