Poster Paper: Leaks in the STEM Pipeline: Evidence from a New Survey of Workers

Saturday, November 10, 2018
Exhibit Hall C - Exhibit Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Drew M. Anderson, Matthew D. Baird and Robert Bozick, RAND Corporation


Training in science, technology, engineering, and mathematics (STEM) is among the most difficult, but also among the most profitable, choices in postsecondary education. However, a large fraction of workers who hold college degrees in STEM do not work in STEM jobs. These individuals paid a high cost without reaping the high benefit.

There are several potential explanations stemming from a mismatch between workers’ skills and the tasks at STEM jobs, the environment or culture of STEM jobs, the benefits and pay at STEM jobs, or the networks required to secure STEM jobs. Existing data sources have not been satisfactory in decomposing and comparing the importance of these explanations. We investigate STEM education and work using a new survey module we developed, asking individuals about their reasons for choosing among STEM and non-STEM jobs.

The survey was fielded in 2017 to the American Life Panel, a nationally representative sample of over 4,000 adults. Focusing on respondents who were of working age (65 or under) and were in the labor force (either employed or seeking employment) we obtained survey responses from 765 workers with STEM degrees. Of them, 60% reported working in STEM jobs. In providing reasons for their job choices, workers could check any of a number of given reasons or write in their own.

The most common single reason STEM-trained workers left STEM was personal preference for the tasks involved in other jobs (32%). However being drawn away by higher demand in non-STEM jobs, either through higher compensation (23%) or more fitting openings (21%), together outnumbered the choices based preference for tasks. Failing to be hired in a STEM job was another prominent reason (22%). Social factors like discrimination or lacking network connections to land a STEM job were uncommon.

Next we explored how these reasons differed across demographic groups. Relative to less-educated workers, those with a bachelor’s degree or higher were much more likely to choose tasks they preferred or that fit better with their training, and much less likely to leave STEM because they were not hired. Older workers were much more likely than younger workers to leave STEM fields for better pay and benefits elsewhere.

Men were more likely than women to leave STEM because they were not hired or could not find a fit for their training. This result is partially explained by our unique subjective definition of STEM, wherein women both classify more jobs as STEM and work in a broader range of STEM jobs, making it less likely they would struggle to find a fit. Given the relatively few responses related to social fit or social networks, we cannot detect evidence of gender discrimination from differences in men and women’s reported reasons for leaving STEM.

This evidence suggests that a large proportion of leaks out of the STEM college-to-career pipeline come from idiosyncratic preferences, with another large proportion coming from market factors drawing workers away. Our results suggest there is not a general shortage of STEM workers, though there are key areas of mismatch.