The replicability and generalizability of internalizing symptom networks across five samples.

Abstract

The popularity of network analysis in psychopathology research has increased exponentially in recent years. Yet, little research has examined the replicability of cross-sectional psychopathology network models, and those that have used single items for symptoms rather than multi-item scales. The present study therefore examined the replicability and generalizability of regularized partial correlation networks of internalizing symptoms within and across five samples (total N = 2,573) using the Inventory for Depression and Anxiety Symptoms, a factor analytically-derived measure of individual internalizing symptoms. As different metrics may yield different conclusions about the replicability of network parameters, we examined both global and specific metrics of similarity between networks. Correlations within and between nonclinical samples suggested considerable global similarities in network structure (rss = .53-.87) and centrality strength (rss = .37-.86), but weaker similarities in network structure (rss = .36-.66) and centrality (rss = .04-.54) between clinical and nonclinical samples. Global strength (i.e., connectivity) did not significantly differ across all five networks and few edges (0-5.5%) significantly differed between networks. Specific metrics of similarity indicated that, on average, approximately 80% of edges were consistently estimated within and between all five samples.The most central symptom (i.e., dysphoria) was consistent within and across samples, but there were few other matches in centrality rank-order. In sum, there were considerable similarities in network structure, the presence and sign of individual edges, and the most central symptom within and across internalizing symptom networks estimated from nonclinical samples, but global metrics suggested network structure and symptom centrality had weak to moderate generalizability from nonclinical to clinical samples. the overall structure, global strength, or individual edges of the networks for the split-halves.Edge weights (rss = .87 and .78), node predictabilities (rss = .93 and .94), and centrality strength (rss = .85 and .86) were consistently strongly correlated between split-halves.Specific Metrics. We examined whether estimated edges replicated across split-halves for each sample, and found that a median of 84.8-89.9% of nonzero edges in the sparser splithalf network replicated (i.e., were nonzero and had the same sign) in the denser split-half network. Examination of consistencies in absent edges revealed that 58.0-69.2% of the absent edges in the denser split-half network replicated in the sparser split-half network. When examining matches in centrality rank-order, dysphoria was identified as the most central symptom in all split-halves. Across all eleven symptoms, however, exact matches in centrality strength rank-order between split-halves were much less frequent (36.4% and 45.5%). Cross-Sample ComparisonsGlobal Metrics. Global characteristics of the five networks are summarized in Table 3.Consistent with prior research showing...

Network analysis of impulse dyscontrol in mild cognitive impairment and subjective cognitive decline

“…Our study could have established more precise estimates with a larger sample size. However, we had relatively few edges to estimate (theoretical maximum of 66), and sample sizes of roughly 200 have been used in similar situations (e.g., Funkhouser et al, 2020). It is also worth noting that this is one of the largest clinical samples of MBI-C data.…”

“…Following the formidable number of exploratory network studies conducted in psychiatric domains, the generalizability and replicability of psychological networks has become a natural frontier for future studies (e.g., Funkhouser et al, 2020;Fried et al, 2018). The generalizability of network structures across clinical and non-clinical domains has the potential to offer tentative mechanistic insights into psychopathology.…”

Examining the relationships between emotional disorder symptoms in a mixed sample of community adults and patients: A network analysis perspective

“…McNally et al (2017) have also shown the lack of connection between the symptoms linked to depression and the symptoms of obsessive-compulsive disorder (OCD), which appear solely connected through "sadness" but not through sleep problems. The results of previous work have also identified dysphoria and lassitude to be the most influential symptoms in the networks obtained for depression (e.g., Funkhouser et al, 2020), whilst traumatic intrusions are the most influential for PTSD (Contractor et al, 2020;Gilbar, 2020) and the checking and ordering symptoms for OCD networks (Zhang et al, 2019).…”

“…In this regard, the Inventory of Depression and Anxiety Symptoms (IDAS; Watson et al, 2007) and the IDAS-II (Watson et al, 2012) have been identified as effective tools for measuring several symptoms and spectra of the transdiagnostic models of psychopathology (Kotov et al, 2017). Funkhouser et al (2020), found consistent connections between the symptoms of dysphoria and lassitude, as well as between dysphoria and social anxiety, evaluated with the IDAS, although this work did not specifically aim to analyze the bridge symptoms that could explain comorbidity. The analysis of the structure of IDAS-II could further provide a broader perspective on the configuration of the relationships between emotional disorders, as it additionally measures symptoms of bipolar disorder (BD) and OCD as well as contains more expansive coverage of PTSD and social anxiety symptoms.…”

Conceptualizing the UPPS‐P model of impulsive personality through network analysis: Key dimensions and general robustness across young adulthood

“…Thus, although results suggested robustness in the general structure of networks across the three waves, further work is needed examining the robustness/sensitivity of specific features within the network (e.g., differences in individual but important edges, statistical differences in dimensions' centrality over time). The debate on network replicability remains ongoing, with additional work needed to determine whether networks may be robust across samples (and whether it is important that they are; Borsboom et al, 2017;Forbes et al, 2017;Forbes, Wright, Markon, & Krueger, 2019;Funkhouser et al, 2020;Robinaugh, Hoekstra, Toner, & Borsboom, 2020).…”