Data analyses were conducted using IBM SPSS Statistics for Windows, Version 23.0 [45 ]. Descriptive statistics were computed to summarize participant characteristics and the dimensions of the Psychological General Well-Being Index (PGWBI), and were presented as means and standard deviations or frequencies and percentages, as appropriate.
Moreover, given the normal distribution of the variables, the relationships between the six dimensions of the PGWBI (anxiety, depression, positive well-being, self-control, general health, and vitality) were examined using Pearson’s correlation coefficients and interpreted as weak (r = 0.1–0.3), moderate (r = 0.3–0.5), or strong (r > 0.5) [46 ]. The strength and direction of the correlations were analyzed to identify significant associations between dimensions and corresponding p-values were calculated. A significance level of p < 0.05 was used for all statistical tests.
In addition, a psychometric network analysis was performed using JASP (Version 0.19.0) [47 ] to disentangle the relationships among the six dimensions of the PGWBI. The analysis was performed using a Gaussian Graphical Model (GGM) with pairwise partial correlations representing the edges (relationships) between nodes (PGWBI dimensions) [48 ]. To minimize the presence of spurious connections, the network model was estimated using the Graphical Least Absolute Shrinkage and Selection Operator (GLASSO) regularization algorithm [49 (link),50 (link),51 (link)].
This approach constrains low correlation values to zero, resulting in a sparse network, by eliminating likely spurious connections. The GLASSO algorithm employs a tuning parameter (λ) to control the sparsity of the network, where higher λ values lead to greater sparsity [49 (link),50 (link),52 ]. Then, the Extended Bayesian Information Criterion (EBIC) was employed as a model-selection criterion to identify and retrieve the most optimal network structure. A γ hyperparameter was set to 0.5 to balance sensitivity and specificity in edge detection [53 (link)].
This EBIC–GLASSO approach has been recognized for its effectiveness in accurately reconstructing true network structures [54 (link),55 (link)] in cases where the network is inherently sparse (i.e., contains relatively few connections). This method also demonstrates high specificity, effectively preventing the estimation of non-existent edges, though its sensitivity (i.e., accuracy in detecting existing connections) can vary.
Furthermore, the stability of the network model was assessed [56 (link)] using the correlation stability coefficient (CS-coefficient). CS-coefficient values higher or equal to 0.5 indicate optimal stability and values higher than 0.25 indicate moderate stability [56 (link),57 (link)].
In addition, centrality measures were computed, including strength, closeness, betweenness, and expected influence. Strength centrality indicates the number of edges (relationships) connected to a node. Closeness centrality measures the proximity of a node to all other nodes, reflecting its level of accessibility within the network. Betweenness centrality measures interactions between nodes, depending on the other nodes that lie on the same path [58 (link),59 (link),60 (link),61 (link),62 (link)]. Last, expected influence accounts for the sum of all positive and negative connections of a node, providing insight into its overall impact on the network.
Moreover, given the normal distribution of the variables, the relationships between the six dimensions of the PGWBI (anxiety, depression, positive well-being, self-control, general health, and vitality) were examined using Pearson’s correlation coefficients and interpreted as weak (r = 0.1–0.3), moderate (r = 0.3–0.5), or strong (r > 0.5) [46 ]. The strength and direction of the correlations were analyzed to identify significant associations between dimensions and corresponding p-values were calculated. A significance level of p < 0.05 was used for all statistical tests.
In addition, a psychometric network analysis was performed using JASP (Version 0.19.0) [47 ] to disentangle the relationships among the six dimensions of the PGWBI. The analysis was performed using a Gaussian Graphical Model (GGM) with pairwise partial correlations representing the edges (relationships) between nodes (PGWBI dimensions) [48 ]. To minimize the presence of spurious connections, the network model was estimated using the Graphical Least Absolute Shrinkage and Selection Operator (GLASSO) regularization algorithm [49 (link),50 (link),51 (link)].
This approach constrains low correlation values to zero, resulting in a sparse network, by eliminating likely spurious connections. The GLASSO algorithm employs a tuning parameter (λ) to control the sparsity of the network, where higher λ values lead to greater sparsity [49 (link),50 (link),52 ]. Then, the Extended Bayesian Information Criterion (EBIC) was employed as a model-selection criterion to identify and retrieve the most optimal network structure. A γ hyperparameter was set to 0.5 to balance sensitivity and specificity in edge detection [53 (link)].
This EBIC–GLASSO approach has been recognized for its effectiveness in accurately reconstructing true network structures [54 (link),55 (link)] in cases where the network is inherently sparse (i.e., contains relatively few connections). This method also demonstrates high specificity, effectively preventing the estimation of non-existent edges, though its sensitivity (i.e., accuracy in detecting existing connections) can vary.
Furthermore, the stability of the network model was assessed [56 (link)] using the correlation stability coefficient (CS-coefficient). CS-coefficient values higher or equal to 0.5 indicate optimal stability and values higher than 0.25 indicate moderate stability [56 (link),57 (link)].
In addition, centrality measures were computed, including strength, closeness, betweenness, and expected influence. Strength centrality indicates the number of edges (relationships) connected to a node. Closeness centrality measures the proximity of a node to all other nodes, reflecting its level of accessibility within the network. Betweenness centrality measures interactions between nodes, depending on the other nodes that lie on the same path [58 (link),59 (link),60 (link),61 (link),62 (link)]. Last, expected influence accounts for the sum of all positive and negative connections of a node, providing insight into its overall impact on the network.
