Clinical Interpretations of Symptom Networks Derived from Cross-Sectional Data: A Critical Re-evaluation Open Access
Hossein, Shabnam (Fall 2018)
The network approach to the study of psychopathology has gained significant interest among clinical psychologists in recent years; largely due to its potential to reveal the temporal dynamics among symptoms. This temporal inference has important implications for the conceptualization of mental disorders, the elucidation of potential “causal” relationships among symptoms, as well as the personalization of clinical interventions. Based on the type of data (longitudinal vs. cross-sectional), one can build directed or undirected symptom networks, both of which offer some degree of insight into the temporal dynamics of symptom relationships. While longitudinal data is preferable in this regard, most of the current literature used cross-sectional data to derive cross-sectional networks, and it is assumed that these networks are at least somewhat generalizable to temporal networks. Consequently, cross-sectional networks have frequently been used as the basis for suggesting that network analysis may help identify the best targets for intervention. The goal of this study was to assess the extent to which properties of symptom networks derived from cross-sectional datasets can be used to make inferences about symptom networks from longitudinal datasets. In study 1, we assessed effects of assumption violations related to the definitions of commonly used network centrality parameters in the psychopathology literature. Here, we used two publicly available cross-sectional datasets to compare different definitions of centrality. We find that depending on the type of network flow assumed in the definitions of centrality, different symptoms might be labeled as the most influential symptoms of the network. In Study 2, we used three separate longitudinal samples of individuals with varying degrees of depressive and anxiety symptoms to compare network properties between cross-sectional and temporal networks. We find that most features of temporal networks cannot be reliably measured based on cross-sectional methods, thereby challenging widely used conventions in the interpretation of cross-sectional symptom networks. Taken together, the results of these two studies suggest that greater caution is warranted in the interpretation of cross-sectional networks, and that more work needs to be done to empirically validate some of the methodological assumptions of the network approach.
Table of Contents
About this Master's Thesis
|Subfield / Discipline|
|Committee Chair / Thesis Advisor|
|Clinical Interpretations of Symptom Networks Derived from Cross-Sectional Data: A Critical Re-evaluation ()||2018-07-30 10:46:35 -0400||