Individual and gender inequality in computer science: A career study of cohorts from 1970 to 2000

TL;DR

It is found that individual inequality in productivity (publications) increases over a scholar’s career but is historically invariant, whereas individual inequality in impact (citations), albeit larger, is stable across cohorts and careers.

Abstract

Inequality prevails in science. Individual inequality means that most perish quickly and only a few are successful, while gender inequality implies that there are differences in achievements for women and men. Using large-scale bibliographic data and following a computational approach, we study the evolution of individual and gender inequality for cohorts from 1970 to 2000 in the whole field of computer science as it grows and becomes a team-based science. We find that individual inequality in productivity (publications) increases over a scholar's career but is historically invariant, while individual inequality in impact (citations), albeit larger, is stable across cohorts and careers. Gender inequality prevails regarding productivity, but there is no evidence for differences in impact. The Matthew Effect is shown to accumulate advantages to early achievements and to become stronger over the decades, indicating the rise of a "publish or perish" imperative. Only some authors manage to reap the benefits that publishing in teams promises. The Matthew Effect then amplifies initial differences and propagates the gender gap. Women continue to fall behind because they continue to be at a higher risk of dropping out for reasons that have nothing to do with early-career achievements or social support. Our findings suggest that mentoring programs for women to improve their social-networking skills can help to reduce gender inequality.

Figures (7)

• Figure 1: Description of the field of computer science. (A) The size of cohorts increases exponentially with time for both males and females. (B) The average team size, measured by the number of authors per paper, increases over time. (C) Distributions of productivity (cumulative number of papers $P$ per author at career age 15) and impact (cumulative number of citations $C$ per author at career age 15) are broad. The lines are best fits to the data: a truncated power law ($P(15)$) and a stretched exponential ($C(15)$). (D) The number of authors decreases with the number of years during which they publish persistently after the beginning of their careers (early career persistence). Female scientists show equal persistence in their early careers but after 4 years they are less likely to persist. (E) The fraction of authors in a cohort that drop out of academia (for ten years in a row) decreases but is more or less constant since the mid-80s. Females drop out more than men.
• Figure 2: Individual inequality in productivity and impact as a function of career ages, depicted for seven cohorts between 1970 and 2000. We count publications and citations cumulatively ($P(t)$ and $C(t)$, defined in "Materials and methods: Individual inequality"). (First two columns) Assigning publications to all authors. (Last two columns) Assigning publications only to first authors. (Second row) Authors are filtered that have not published for ten consecutive years (most likely left academia).
• Figure 3: Individual inequality in productivity and impact as a function of cohorts, depicted for career ages 3, 5, 10, and 15. We count publications and citations cumulatively ($P(t)$ and $C(t)$, defined in "Materials and methods: Individual inequality"). (First two columns) Assigning publications to all authors. (Last two columns) Assigning publications only to first authors. (Second row) Authors are filtered that have not published for ten consecutive years (most likely left academia).
• Figure 4: Gender inequality for productivity and impact as a function of cohort and career ages. We compare the cumulative publications distribution $P_{\mathrm{gender}}(t)$ and cumulative citations distribution $C_{\mathrm{gender}}(t)$ of male and female scientists in the same cohort at the same career age $t$ and test differences between these distributions. Color marks the effect size (Cliff's $d$). Positive values (red) indicate that men dominate women, while negative values (blue) reveal that women dominate men. Effects are only shown if they are significant ($p\leq0.05$) according to a Mann–Whitney $U$ test. Details in "Materials and methods: Gender inequality." Publications are assigned to all authors (A, B) or first authors only (C, D). In general, effects decrease with cohort and increase with career age.
• Figure 5: Matthew Effect. (First column: A, F) Measurement of the strength of a cohort's reproductive feedback as the exponent that relates an author's number of papers produced, or citations received, in a career age (y-axis) to the respective cumulative numbers in the previous career age (x-axis), shown for the 2000 cohort and the last career age. Exponents show as slopes of the continuous lines on the log-log plot. Dotted lines indicate that feedback fully unfolds only above a lower cutoff. (B, G) For an average cohort, potential individual advantages from feedback are constant along the career path, for both productivity and impact. (C, H) For an average cohort, the number of citations required to take advantage of feedback increases along the career path. (D, I) For an average career age, potential individual advantages from feedback increase historically, but more so for productivity. (E, J) For an average career age, the numbers of citations and publications required to take advantage of feedback increase historically. (All columns but the first) Shaded areas are bounded by minima and maxima, lines show means.