*2.4. Sample Size Considerations and Network Stability/Edge Accuracy Calculations*

Currently, there is no established method for a formal power analysis available for psychological network analyses, nor is there a minimal sample size required for this type of analysis [43]. Rather, there are established methods to evaluate the network stability (e.g., stability of central indices) and edge accuracy which should be reported along a psychological network analysis. Providing evidence for the stability of a network solution is an important prerequisite to reasonably interpret the network. In general, there is a larger chance to find stable network solutions in larger samples than in smaller samples.

The accuracy of the network solution was evaluated by the two following analyses: First, the case-dropping subset bootstrap approach was used to analyze the stability of

central indices after observing only subsamples of the data. The correlation stability (CS) coefficient quantifies the stability of central indices and represents the maximum proportion of cases that can be dropped from the full dataset so that the correlation between the original central indices and central indices calculated from bootstrap subsets has a 95% probability of being *r* = 0.7 or higher. Ideally, the CS-coefficients should be above 0.5 [43]. Second, the accuracy of edge weights was evaluated by calculating 95% confidence intervals based on non-parametric bootstrapping (*n* = 1.000 boots) which is recommended for LASSO regularized edges [43]. In case of excessively large bootstrapped confidence intervals, the edge strengths should be interpreted with caution.
