Figure 3

Competitive arenas in science. The right-skewed distributions of scientist longevity L and productivity $N_p$ are the starting point for understanding the skewed success distributions in science. Panels A (Economics) and B (Nat./PNAS/Sci.) address censoring bias in the longevity distributions by considering separately the cohorts of scientists who entered the Nat./PNAS/Sci. arena in the indicated time window. Panel C (Nat./PNAS/Sci.) shows the unconditional aggregate longevity distribution. Panels D/E/F show $P(\ge N_p)$, the complementary cumulative distribution function (CCDF) of total publications $N_p$. Panels D/E are analogs of panels A/B, showing the fixed cohorts of researchers who entered the respective journal arena with their first publication in the indicated time window. Panel F shows the $P(\ge N_p|L)$ distributions conditioned on L, which indicate that even within the set of ‘iron horse’ researchers with longevity $L_i\ge 11$ the distribution is still extremely broad. To guide the eye and emphasize the curvature in $P(\ge N_p)$, which is not consistent with a power-law distribution, we plotted a curve proportional to $N_p^{-2}$.