The case of computer science education, employment, gender, and race/ethnicity in Silicon Valley, 1980–2015

ABSTRACT

We analyse race and gender trends in the Silicon Valley technology industry from 1980 to 2015, with a focus on education, employment and wages in computer science. Racial gaps in representation are more salient among programmers than in the overall technology labour force; in addition, we document a stable or increasing gender gap across all races in computer science. However, these demographic shifts are not always consistent with either a pipeline argument that there are insufficient supplies of potential underrepresented programmers or a wage explanation. Hispanic males, for example, have had increasing rates of computer science degree completions yet decreasing representation in the programmer labour force. On the other hand, White females have had decreasing representation among both degrees and the labour force despite comparatively high wages in the technology sector. The persistent and increasing race and gender gaps suggest that policies to attract underrepresented groups need to be differentiated by the group and may require significant changes in industry culture to increase the representation of these groups.

KEYWORDS:

• Race
• gender
• higher education
• computer science
• technology

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Burke and Mathis, 2007; Marginson et al. 2013. For the U.S., see, National Academy of Sciences, 2007, and Xie, Fang, and Shauman (2015). UNESCO has also focussed on gender gaps in access to STEM jobs and job mobility. http://www.unesco.org/new/en/natural-sciences/priority-areas/gender-and-science/improving-measurement-of-gender-equality-in-stem.

2. Open Doors is a comprehensive report on international students studying in the U.S. and U.S. students studying abroad, supported by the U.S. Department of State and published by the Institute of International Education.

3. We recognise that the Silicon Valley technology labour market may be a national market, but restrict the higher education analysis to California for comparability.

4. The long form of the population census ceased in 2000.

5. This includes respondents in the following counties: Alameda, Contra Costa, San Francisco, San Mateo, Santa Clara and Santa Cruz. Silicon Valley is not an official government designation and thus we use an inclusive geographic region in our analyses.

6. The National Centre for Education Statistics uses these definitions of full-time and full-year.

7. We use the harmonised occ1990 occupation category of 229 (programmers) which is defined as computer software developers and computer scientists/analysts (occ1990).

8. In the year 2015 census data uses intervalled wage data unlike the other census years, and thus the average of these intervals is used as the wages for 2015.

9. However, changes in the demographics of Silicon Valley’s labour force differed from those in the rest of the country. From 1980 to 2015, Silicon Valley’s White and Black labour force across all industries declined more than nationally, but Hispanics and Asians increased much more rapidly, reaching 50% of all workers, compared to 30% nationwide.

10. We define an industry as belonging to the technology industry if the industry is listed as ‘Computers and related equipment’ (#322), ‘Radio, TV, and communication equipment’ (#341), ‘Electrical machinery, equipment, and supplies, nec’ (#342), ‘Guided missiles, space vehicles, and parts’ (#362), ‘Scientific and controlling instruments’ (#371), ‘Computer and data processing services’ (#732), ‘Engineering, architectural, and surveying services’ (#882) or ‘Research, development, and testing services’ (#891) in the harmonised industry variable (ind1990). Manufacturing industries were industries with the codes 100-392 in the harmonised industry variable (ind1990), excluding those in the computer category. High services industries were industries with the codes 700-712, 721,732, and 812-893 in the harmonised industry variable (ind1990), excluding those in the computer industry.

11. Occupations are categorised as Manager with the codes 004-022 in the harmonised occupation category (occ1990). These do not include management-related occupation such as accountants or HR specialists and include executives (there were too few executives to be a separate category). Occupations are categorised as Professionals with the codes 043-200 (Professional Speciality list), 229 (programmers), and 23-37 (Management-Related occupations) in the harmonised occupation category (occ1990). All other occupations are categorised as ‘Other’ in these analyses (includes occupations such as cook, bookkeeper, waiter, office clerk, etc.).

12. Data tables are available upon request.

13. Authors calculated these percentages from IPEDS data (data available upon request).

14. Authors calculated these percentages from Open Doors data (data available upon request).

15. Data used in analyses are from IPEDS/Open Doors and are available upon request.

16. Detailed data are available upon request.

17. Detailed data are available upon request.

18. Unfortunately, because of the small sample sizes for Hispanics and Blacks, we are forced to restrict the wage analyses to White and Asian full-time full-year workers. We also limit the sample to 25-44-year-olds, include only positive wages, and separate analyses into programmers with undergraduate degrees only and programmers with graduate degrees–this to provide less biased wage comparisons.

19. As mentioned above, non-US citizens may continue to major in CS and accept lower pay than their U.S. citizen counterparts because they face much lower wages should they return home. A more complex question is why US citizen Asian males with undergraduate degrees receive lower wages than their White male counterparts.

Additional information

Notes on contributors

June Park John

June Park John recently finished her PhD at Stanford. Her research interests are economics of education and race & gender differences at the intersection of education and technology.

Martin Carnoy

Martin Carnoy is the Vida Jacks Professor of Education at Stanford University School of Education. Prior to coming to Stanford, he was a Research Associate in Economics, Foreign Policy Division, at the Brookings Institution. He is also a consultant to the World Bank, Inter-American Development Bank, Asian Development Bank, UNESCO, IEA, OECD, UNICEF, International Labour Office.