Table 1.
Covariate means for main regression specifications.
Fig 1.
Composition of Science Graduates in paid and unpaid work 6 months after graduation.
Unpaid workers are disproportionately women with good degrees, regardless of SES class. Note: Sample includes those who completed the survey 6 months after graduation, had complete records of SES and degree class, and were in paid or unpaid work. Stars indicate: * = <0.05; ** = <0.01 for two-sample test of differences in proportions.
Table 2.
Characteristics of science graduates in paid and unpaid work 6 months after graduation.
Fig 2.
Multivariate regression analysis of earnings and persistence in STEM.
Top panels: Conditional marginal differences are compared to ‘Paid Work at 6 months, in STEM’. Relative to paid work not in STEM, whether an unpaid position is in STEM or not is conditionally uncorrelated with salaries, but for paid workers initially taking a STEM role has a persistent association with salaries. Bottom panels: Conditional marginal differences are compared to ‘Paid work at 6 months, not obtained by personal connection’. Taking unpaid work, regardless of how it was obtained, had a negative association with salary. Taking unpaid work only has a negative association with persistence in STEM when not found through personal connections. Note: Error bars represent 95% confidence intervals for the conditional marginal difference.
Table 3.
Multivariate regression of whether six-month position was found through personal connections or was in STEM, on demographic and job characteristics.
Fig 3.
Characteristics of paid and unpaid work among science graduates six months after graduation.
Unpaid work was found more frequently through personal connections than paid work, while a greater proportion of paid workers were in a STEM field.