Table 1.
Summary of the results reported in Fried et al. (2018)
| Network characteristic | Parameter estimate | Interpretation in Fried et al. (2018) |
|---|---|---|
| t-tests between samples for average symptom endorsement | Except for Sample 1 compared to Sample 2, all samples differed significantly in mean symptom severity (ps < 2.2.×10−16) | “Samples differed in average symptom endorsement” (p. 6) |
| Spearman rank correlations among mean symptom profiles | Range from rs = [.34] to rs = .82 (median rs = .63). | “There were considerable similarities across data sets in their mean symptom profiles” (p. 6) |
| Visual comparison of the networks1 | N/A | “The four networks featured many consistent edges…There were also specific edges that differed considerably…” (p. 8). |
| Symptom strength centrality order (Pearson correlations between each pair of networks)1 | Range from r = .63 to r = .75 (median r = .69) | “centrality order was substantially related across the four networks” (p. 8) |
| bootnet results2 | ||
| Correlation-stability coefficients | Range from CS(0.7) = .52 to CS(0.7) = .75 (median CS(0.7) = .60) | “The correlation-stability coefficient[s] for strength centrality…exceeded the recommended threshold for stable estimation of 0.50” (p. 8) |
| Edge weight 95% confidence intervals | Plot of CIs (see Fig S3 in Fried et al.) | “Stability analyses indicated that all four networks were accurately estimated, with small to moderate confidence intervals around the edge weights” (p. 8). |
| Coefficient of similarity (Spearman rank correlations of edge lists)1 | Range from rs = .62 to rs = .74 (median rs = .71) | “indicating strong similarities” (p. 8). |
| NetworkComparisonTest results3 | ||
| Omnibus test of network structure invariance | all ps < .005 | “implying that no pair of networks featured exactly the same 120 edge weights” (p. 8) |
| Individual edge invariance test | Range of 2 (1.7%) to 8 (6.7%) edges differed significantly across the six comparisons (median of 3 [2.5%]) | “Overall, networks were moderately to strongly correlated and only a few significantly different edges emerged, which implies considerable similarities.” (p. 9) |
| Global strength invariance test | Four of the six pairwise comparisons were significantly different (p < .05) | “Global strength values were fairly similar” (p. 9) |
| Cross-sample variability network4 | Three edges were described as “the most variable”, with standard deviations of.15, .15, and .14 | “For the remaining edges, standard deviations were small to negligible” (p. 10) |
| Summary of the results | N/A | “First, whereas data sets differed in overall PTSD severity, the patterns of symptom endorsement were correlated across the four samples… Second, whereas the structures of the four networks were not statistically identical (i.e., not all edges were exactly the same), the networks showed moderate to high intercorrelations, as did strength centrality coefficients. Third, we highlighted the most pronounced differences among networks by estimating a variability network: [Three edges] differed considerably across the four samples, whereas other edges were similar or identical across networks.” (p. 10) |
Note.
1
Based on jointly estimated Gaussian graphical models (GGMs) derived from polychoric correlations using fused graphical lasso selecting tuning parameters using k-fold cross-validation.
2
Based on individually estimated GGMs derived from polychoric correlations.
3
Based on individually estimated GGMs derived from Pearson correlations, excluding the 0.3–3.8% of cases with missing data.
4
Based on jointly estimated Gaussian graphical models (GGMs) derived from polychoric correlations using fused graphical lasso selecting tuning parameters using information criteria.