Gender shapes the relationship between productivity and journal prestige in science

Abstract

Gender disparities in academia manifest and persist in various aspects of the scientific enterprise, yet their influence on the interplay between research productivity and journal prestige remains underexplored. Here we analyze the academic trajectories of over 6,000 elite Brazilian researchers by jointly tracking their annual productivity and the average prestige of the journals in which they publish. By projecting individual career years onto a standardized productivity-prestige plane and applying Bayesian hierarchical modeling, we find that male researchers are more likely to follow productivity-oriented trajectories and are markedly overrepresented in the hyperprolific region of this plane. Female peers, in contrast, more often occupy regions that prioritize journal prestige over publication quantity. Although male researchers publish more throughout their careers, their female counterparts achieve comparable or higher average journal prestige, particularly in later career stages and among outlier individuals. Male researchers also exhibit greater temporal persistence in their productivity and impact levels and are especially averse to simultaneously changing both metrics compared to their female peers. Among non-outliers, productivity and career age have a negative overall impact on the average journal prestige of researchers of both genders, with slightly stronger effects observed among female researchers; however, these patterns vary across disciplines, highlighting the complexity and heterogeneity of academic careers.

Figures (5)

• Figure 1: Gendered patterns in the journal prestige-productivity plane. (a) Scatter plot showing the relationship between yearly average journal prestige and productivity across career years of researchers in our dataset. Each point represents a career year of a male (purple) or female (green) researcher, with both variables expressed in standard score units. One unit of productivity corresponds to a one standard deviation above (positive) or below (negative) the average productivity within the researcher's discipline and year. Similarly, one unit of journal prestige represents one standard deviation above (positive) or below (negative) the expected value given the same productivity level within the researcher's discipline and year. The plane is divided into four non-outlier sectors ($I\!\!+\!\!P+$, $I\!\!+\!\!P-$, $I\!\!-\!\!P+$, and $I\!\!-\!\!P-$), based on whether metrics are above or below the disciplinary averages, and three outlier sectors ($I\!P\!\!+\!\!+$, $P\!\!+\!\!+$, and $I\!\!+\!\!+$) corresponding to exceptionally high values of journal prestige and/or productivity. The inset depicts the full extent of the plane. (b) Venn diagrams illustrating the set relationships of our classification of researchers into non-outliers (no career year in an outlier sector), perfectionists (at least one career year in $I\!\!+\!\!+$), hyperprolifics (at least one career year in $P\!\!+\!\!+$), and hyperprolific perfectionists (at least one career year in $I\!P\!\!+\!\!+$). The upper diagram displays the results for male researchers, while the lower one shows the results for female researchers. (c) Probability of finding male (purple) and female (green) perfectionist researchers as a function of the number of hyperprolific years in their careers, as estimated via a logistic regression model. The inset shows the fitted logistic coefficients.
• Figure 2: Gendered patterns in the occupation of the journal prestige-productivity plane. (a) Occupation fraction of career years of male (purple) and female (green) researchers in different sectors of the plane, compared with the overall gender prevalence of career years (first bar and dashed line). Numbers above the bars indicate the total percentage of career years per sector, while numbers within the bars denote the gender-specific occupation fractions. (b) Excess of career years of male (purple) and female (green) researchers in each sector relative to a null model in which gender is randomly shuffled across all career years. Error bars indicate 95% confidence intervals from 1,000 realizations of the null model. (c) Productivity gap, defined as the difference between the average productivity of female and male researchers within each sector. (d) Journal prestige gap, defined as the difference between the average journal prestige of female and male researchers within each sector. Error bars in panels (c) and (d) represent 95% bootstrap confidence intervals.
• Figure 3: Gender differences in transitions between sectors of the journal prestige-productivity plane. Standardized transition frequencies between all pairs of sectors observed over the careers of male (left panel) and female (right panel) researchers, further grouped into outliers and non-outliers researchers. For each possible transition and gender, the values represent $z$-scores calculated by subtracting the observed transition frequency from the average frequency obtained via randomly shuffling researchers' trajectories and dividing by the standard deviation of the corresponding shuffled outcomes.
• Figure 4: Evolution of gendered patterns over researchers' careers. (a) Occupation of sectors of the journal prestige-productivity plane at five-year intervals over the careers of male (left) and female (right) researchers. (b) Evolution of average productivity (left) and average journal prestige (right) over the careers of male (purple) and female (green) non-outlier researchers. (c) Evolution of average productivity (left) and average journal prestige (right) over the careers of male (purple) and female (green) outlier researchers. Shaded areas in panels (b) and (c) represent 95% bootstrap confidence intervals, and dashed lines indicate the zero baseline.
• Figure 5: Role of gender in the effects of productivity and career age on average journal prestige of non-outlier researchers. Posterior probability distributions of the population-level parameters describing (a) baselines for journal prestige ($\mu_\alpha$ for males and $\mu_\alpha + \tilde{\mu}_\alpha$ for females), (b) the effect of productivity ($\mu_P$ for males and $\mu_P + \tilde{\mu}_P$ for females), and (c) the effect of career age ($\mu_A$ for males and $\mu_A + \tilde{\mu}_A$ for females). The top row presents results for all non-outlier researchers, while subsequent rows correspond to each of the 14 disciplines in the JIF dataset. Distributions for male researchers are shown in purple, and those for female researchers in green. Vertical dashed lines denote the zero baseline.