Table 4 Measures of centrality robustness.
Forum A | Forum B | |||||
|---|---|---|---|---|---|---|
Measure | Description | Comparison network | In-degree | Out-degree | In-degree | Out-degree |
Top 1 | Is the most central node in the public replies layer the same as the most central node in the private messaging layer/the flattened network? | Private messaging | No | No | Yes | Yes |
Flattened | Yes | Yes | Yes | Yes | ||
Top 3 | Is the most central node in the public replies layer in the top 3 of the most central nodes in the private messaging layer/the flattened network? | Private messaging | No | No | Yes | Yes |
Flattened | Yes | Yes | Yes | Yes | ||
Top 10% | Is the most central node in the public replies layer among the 10% most central nodes in the private messaging layer/the flattened network? | Private messaging | Yes | Yes | Yes | Yes |
Flattened | Yes | Yes | Yes | Yes | ||
Overlap (Szymkiewicz–Simpson coefficient) | What is the overlap between the 10% most central nodes in the public replies layer and the private messaging layer/the flattened network? | Private messaging | 0.74 | 0.60 | 0.93 | 0.88 |
Flattened | 1 | 1 | 0.98 | 0.97 | ||
R | What is the correlation of the centralities between the public replies layer and the private messaging layer/flattened network (limited to actors who are in both layers)? | Private messaging | 0.54 | 0.39 | 0.63 | 0.63 |
Flattened | 0.93 | 0.94 | 0.98 | 0.998 | ||
