Auditing the research practices and statistical analyses of group-level temporal network approach to psychological constructs: A systematic scoping review

Abstract

Network analyses have become increasingly common within the field of psychology, and temporal network analyses, in particular, are quickly gaining traction, with many of the initial articles earning substantial interest. However, substantial heterogeneity exists within the study designs and methodology, rendering it difficult to form a comprehensive view of its application in psychology research. Since the field is quickly growing and that there have been many study-to-study variations in terms of choices made by researchers when collecting, processing, and analyzing data, we saw the need to audit this field and formulate a comprehensive view of current temporal network. To systematically chart researchers’ practices when conducting temporal network analyses, we reviewed articles conducting temporal network analyses on psychological variables (published until March 2021) in the framework of a scoping review. We identified 43 articles and presented the detailed results of how researchers are currently conducting their temporal network analyses. A commonality across results concerns the wide variety of data collection and analytical practices, along with a lack of consistency between articles about what is reported. We use these results, along with relevant literature from the fields of ecological momentary assessment and network analysis, to formulate recommendations on what type of data is suited for temporal network analyses as well as optimal methods to preprocess and analyze data. As the field is new, we also discuss key future steps to help usher the field’s progress forward and offer a reporting checklist to help researchers navigate conducting and reporting temporal network analyses.

On the Measurement of Climate Change Anxiety: French Validation of the Climate Anxiety Scale

“…As such, a critical next step in future iterations would thus be to further establish the theoretical and clinical relevance of distinguishing these two features of climate anxiety. Moreover, since psychological mechanisms fluctuates over time (e.g., Blanchard et al, 2022 ; Heeren et al, 2015 ), it makes sense to also look at their temporal variations. In the same vein, one may wonder about their temporal unfolding.…”

Parental burnout features & their family context: A temporal network approach in mothers

“…A first limitation is that the residuals were not normally distributed, which is an assumption for the multilevel VAR model. In a recent scoping review about temporal network analyses, less than a quarter of studies examined whether the assumption of normality is violated (Blanchard, Contreras, et al, 2022), and little is known about how nonnormal data, or transforming the data, might affect the results or interpretation-although hopefully this will be a target for future statistical and theoretical development. Therefore, we chose to report the raw data in this manuscript and the transformed networks in the supplementary materials as a sensitivity analysis, similarly to Faelens et al (2021).…”

“…As preregistered, we conducted a sensitivity analysis to examine whether transforming data to adhere to a normal distribution would change the pattern of results; it did (specifically for the temporal network, which was sparser with transformed data; see the supplementary materials for more details). As there is little information on how transforming non-normal intensive longitudinal data could impact the interpretation of temporal network analyses (Blanchard, Contreras, et al, 2022), we report network analyses based on the raw results in the manuscript. However, we report and discuss a model estimated from the transformed data in the supplementary materials (see Figure S3).…”

The network structure of psychopathological and resilient responses to the coronavirus pandemic: A multi-country study of general population depression and anxiety symptoms

“…Moreover, we adopted a dynamic perspective by considering two timepoints to create the psychological response groups, however, the data modelled in the network are cross-sectional (from T1) and thus, precludes inferences about causal relationships being drawn. Future research would benefit from adding more time points assessments, or utilising intensive longitudinal data, which would allow the application of, for instance, panel data analysis (Mertens et al, 2017) or temporal network approach (Blanchard et al, 2022) respectively, that could reveal temporal prediction closer to causality.…”