The replicability and generalizability of internalizing symptom networks across five samples

Area of Science:

• Psychopathology research
• Network analysis
• Quantitative psychology

Background:

• Network analysis is increasingly popular in psychopathology research.
• Replicability of psychopathology network models, especially using multi-item symptom scales, is understudied.
• Previous research often relied on single-item symptom measures.

Purpose of the Study:

• To examine the replicability and generalizability of network models for internalizing symptoms.
• To compare network structures and centrality using multi-item symptom scales across clinical and nonclinical samples.
• To assess network similarity using both global and specific metrics.

Main Methods:

• Utilized regularized partial correlation networks for internalizing symptoms.
• Analyzed 5 independent samples (N=2,573) using the Inventory for Depression and Anxiety Symptoms.
• Applied global and specific similarity metrics to compare network parameters.

Main Results:

• High global similarity in network structure (rS=0.53-0.87) and centrality (rS=0.37-0.86) within nonclinical samples.
• Weaker similarity in network structure (rS=0.36-0.66) and centrality (rS=0.04-0.54) between clinical and nonclinical samples.
• Approximately 80% of edges were consistently estimated; dysphoria was the most central symptom across samples.

Conclusions:

• Internalizing symptom networks show considerable within-group similarity in nonclinical samples.
• Network structure and symptom centrality generalize moderately from nonclinical to clinical populations.
• Findings highlight the importance of using multi-item scales for robust network analysis in psychopathology.