Higher publication impact of female faculty is correlated with higher relative risk of academic career choice.

<p>Risk in academic career choice and difference in publication impact. We quantify the risk of academic career choice according to Eq. (10). We show results for two alternative measures of difference in publication impact. In (<b>A</b>), we defined the gender difference in publication impact as the average <i>h</i>-index z-scores of females. The error bars indicate standard errors. See Fig. S13 for the statistical significance of the gender difference in publication impact. The trend line (black dashed line) indicates a significant positive correlation (coefficient of determination ). In (<b>B</b>), we defined the gender difference in publication impact as the probability that female authors have larger <i>h</i>-index z-scores than male authors, as depicted in Fig. S13. The trend line (black dashed line) indicates a significant positive correlation (coefficient of determination ). Note that the values of the risk of academic career choice in (<b>A</b>) and (<b>B</b>) are different for each discipline because the coefficients in the linear regression are different. The data suggest that in disciplines where it is risky to pursue an academic career, female faculty have publications with higher impact than male faculty.