Dear Johann,
I changed initial link threshold and got different graph components.
Here are another questions.
I changed also some parameters as I checked in an attached figure.
According to various parameters, I got totally same graph component but different alpha value in 'Network Inspector'.
When I set the parameters (generate : 1 / Number of repetitions : 1), it showed best alpha value, zero.
However when I set the parameters (generate : 10 / Number of repetitions : more than 10), it showed higher alpha value above 0.5.
Could I trust the result which showed zero alpha value?
Best regards,
Saske
| Attachment | Size |
|---|---|
| 2.JPG | 160.29 KB |
Forums
Reply: Various parameters
Hi,
it seem as if you are using GraphVar the wrong way. If you want to do calculations on the raw matrices, you don´t have to genereate subject specific random networks in the "network construction panel" (please uncheck this box; upper panel).
If you are doing group comparisons on the raw matrices (i.e., non-parametric permutations tests) you need to check "test against random groups" in the "statistics panel" (as you did). The number of repetitions refers to the amount of permutations. I would suggest at least 100 (or even more to come up with a reliable estimation). The initial link threshold (that I previously refered to) will have to be selected in the network inspector later on.
I would strongly suggest that you please study the respective paragraphs in the manual or in the article to the toolbox - otherwise you may come up with invalid results!
Best,
Johann
Thank you for fast reply
Dear Johann,
Thank you for your details.
I will ckeck the documents which are related to GraphVar.
Best,
Saske