I have been working on network analyses to assess DSM-IV PTSD symptomology using CAPS data for my study sample of women who have recently experienced interpersonal violence. I have created partial correlation networks using qgraph and LASSO regularization. The problem I have seems to be a common one- I have a very small sample size (current n = 72). With gammas of 0 to .22, networks with plenty of edges are generating. For gammas of .23 onward, the network is empty. Further, when using the EstimateNetwork function, the best fitting network is also determined to be empty.

My question is whether the network and centrality data for networks with gammas of 0-.22 are legitimate or too compromised due to low power. Noting the limitation of sample size, is that data still something I could disseminate, as in through a conference poster or, if possible, a manuscript?

Thank you!