Abstract
We discuss the Gaussian graphical model (GGM; an undirected network of partial correlation coefficients) and detail its utility as an exploratory data analysis tool. The GGM shows which variables predict oneanother, allows for sparse modeling of covariance structures, and may highlight potential causal relationships between observed variables. We describe the utility in 3 kinds of psychological datasets: datasets in which consecutive cases are assumed independent (e.g., crosssectional data), temporally ordered datasets (e.g., n = 1 time series), and a mixture of the 2 (e.g., n > 1 time series). In timeseries analysis, the GGM can be used to model the residual structure of a vectorautoregression analysis (VAR), also termed graphical VAR. Two network models can then be obtained: a temporal network and a contemporaneous network. When analyzing data from multiple subjects, a GGM can also be formed on the covariance structure of stationary means—the betweensubjects network. We discuss the interpretation of these models and propose estimation methods to obtain these networks, which we implement in the R packages graphicalVAR and mlVAR. The methods are showcased in two empirical examples, and simulation studies on these methods are included in the supplementary materials.
Original language  English 

Pages (fromto)  128 
Journal  Multivariate Behavioral Research 
DOIs  
Publication status  Published  16 Apr 2018 
Keywords
 timeseries analysis
 multilevel modeling
 multivariate analysis
 exploratorydata analysis
 network modeling
Fingerprint Dive into the research topics of 'The Gaussian graphical model in crosssectional and timeseries data'. Together they form a unique fingerprint.
Profiles

Rene Mottus
Person: Academic: Research Active