Table S5 Mean value of network summary statistics for five hiring models vs. the empirical values for Computer Science. Mean values of the three network summary statistics—eigenvector centrality, harmonic centrality, and normalized mean geodesic distance—for networks produced by each of the five hiring models vs. the corresponding values of the empirical Computer Science network. Boldface values are close to the empirical value. Uncertainty indicates the average absolute distance between predicted and empirical value. For this comparison, parameter combinations achieving the closest structural inequality are applied when \(\beta = 1.0\) (Table S3). Two initializations (Init.) of Egalitarian (E) and Skewed (S) are used. Dashed line represents the Ranking and Local homophily model are unable to reproduce realistic structural inequalities. All values are rounded to the second digit
Hiring Model | Init. | Eigenvector centrality (log) (Empirical, −6.7) | Harmonic centrality (Empirical, 0.34) | Geodesic distance (Empirical, 0.34) |
|---|---|---|---|---|
Faculty production | E | −3.85 ± 3.5 | 0.39 ± 0.09 | \(\mathbf{0.34 \pm 0.11}\) |
S | \(\mathbf{-3.89 \pm 3.52}\) | 0.39 ± 0.09 | 0.35 ± 0.12 | |
Local homophily | E | – | – | – |
S | −13.88 ± 8.4 | 0.17 ± 0.21 | 0.34 ± 0.14 | |
Ranking | E | – | – | – |
S | – | – | – | |
Rank-weighted production | E | −1.8 ± 4.91 | 0.45 ± 0.12 | 0.47 ± 0.17 |
S | −1.79 ± 4.91 | 0.45 ± 0.12 | 0.46 ± 0.16 | |
Rank-weighted homophily | E | −1.92 ± 4.79 | \(\mathbf{0.35 \pm 0.13}\) | 0.43 ± 0.15 |
S | −1.96 ± 4.76 | \(\mathbf{0.35 \pm 0.13}\) | 0.44 ± 0.16 |