Fig. 3: Differential attrition exacerbates inequalities in domestic faculty production.
From: Quantifying hierarchy and dynamics in US faculty hiring and retention

a, Line segments contrast the faculty production Gini coefficients calculated for newly hired faculty (filled circles; n = 54,100) and for existing faculty (open circles; n = 184,576) for each of the 107 fields (colours), eight domains (grey) and academia as a whole (blue). Line segments are grouped and coloured by domain. b, Annual Gini coefficients for academia and for each domain showing strong interyear consistency. c, Attrition risk as a function of university production rank by domain and for academia overall, via logistic regression, showing that university production rank is a significant predictor of annual attrition risk (two-sided t-test, Benjamini–Hochberg-corrected P < 0.05) such that faculty trained at high-producing universities leave academia at substantially lower rates than those trained at less productive universities. The empirical average annual attrition rates vary around the fitted curves.