Problems with Centrality Measures in Psychopathology Symptom Networks: Why Network Psychometrics Cannot Escape Psychometric Theory

Abstract

Understanding patterns of symptom co-occurrence is one of the most difficult challenges in psychopathology research. Do symptoms co-occur because of a latent factor, or might they directly and causally influence one another? Motivated by such questions, there has been a surge of interest in network analyses that emphasize the putatively direct relationships among symptoms. In this critical paper, we highlight conceptual and statistical problems with using centrality measures in cross-sectional networks. In particular, common network analyses assume that there are no unmodeled latent variables that confound symptom co-occurrence. In simulations that include latent variables, we demonstrate that closeness and betweenness are vulnerable to spurious covariance among symptoms that connect subgraphs (e.g., diagnoses).Furthermore, strength is redundant with factor loading in several cases. Finally, if a symptom reflects multiple latent causes, centrality metrics reflect a weighted combination, undermining their interpretability in empirical data. Our results suggest that it is essential for network psychometric approaches to examine the evidence for latent variables prior to analyzing or interpreting symptom centrality. Failing to do so risks identifying spurious relationships or failing to detect causally important effects. Altogether, centrality measures do not provide solid ground for understanding the structure of psychopathology when latent confounding exists.

Towards a Network Psychometrics Approach to Assessment: Simulations for Redundancy, Dimensionality, and Loadings

“…The main implication is that my simulation provides further evidence that node strength is statistically equivalent to factor loadings (Hallquist et al, 2019). Despite this equivalence, it's important to consider their substantive interpretations.…”

“…The implications of the redundancy analysis should be far reaching for network analysts in psychology. Identifying unique components of psychological attributes is essential for understanding the processes that underlie them as well as valid measurement of the attribute itself (Hallquist et al, 2019). Assessment instruments in personality, for example, are often redundant, which may make for more reliable measures but may also decrease the validity of the measurement (McCrae & Mõttus, 2019).…”

“…A key part of this definition is that these components are unique in that they are causally autonomous (i.e., distinct causal processes). A recent set of simulation studies corroborated this point by demonstrating that network measures are affected by latent confounding (e.g., similar item phrasings, underlying common causes; Hallquist et al, 2019). Therefore, there is a need to identify unique components in networks to (a) align with the theoretical perspective of a causal system and (b) ensure the accurate interpretation of network measures.…”

“…The second finding suggests that the relationship between node strength and factor loadings should be tempered in a way that reflects the unique latent causes in the system. This latter takeaway jibes with the notion that the unique causal components must be identified before network measures can be meaningfully interpreted (Christensen et al, under review;Hallquist et al, 2019).…”

“…A notable example comes from personality questionnaires, where item content tends to overlap in order to measure a specific attribute (e.g., extraversion). This same attribute from the network perspective is suggested to be comprised of unique causal components, which makes the overlap of item content problematic due to latent confounding Hallquist, Wright, & Molenaar, 2019). Although this may seem to suggest that there is a need to start anew, this is not necessarily the case.…”