Bootstrapping edges after regularization: clarifications & tutorial

I have read the following statement numerous times in the last months: “If the 95% boostrapped CI of an edge weight in a network contains zero, the edge is likely not different from zero”. This is incorrect in case of regularized partial correlation networks, and I thought I would write …

FAQ on network stability, part II: Why is my network unstable?

In a previous tutorial blog post, I summarized the bootstrapping routine developed by Sacha Epskamp and me that we recommend for testing the stability / accuracy / precision of estimated network models and parameters. This matters a great deal and should be considered a standard step in every empirical network …