The field of computer science (CS) is currently experiencing a surge in undergraduate degree and course enrollments. This surge is straining program resources at many institutions, and causing concern among faculty and administrators about how best to respond to the rapidly growing demand. There is also concern about what this will mean for the future of these CS programs, the field as a whole, and U.S. society more broadly. The computing community has recently begun calling attention to and seeking solutions to this challenge.1
The Committee on the Growth of Computer Science Undergraduate Enrollments was asked to examine the recent phenomenon of increasing enrollments in undergraduate computing courses and to assess the scope of this trend and the underlying drivers, likely future enrollment trends, and the impact of recent enrollment growth on diversity in the computing disciplines. The committee’s full Statement of Task is contained in Box 1.1 (and in Appendix A).
The following report is the output of the committee’s work. It explores existing data and evidence about historical and recent trends in computing, distills key findings, identifies actions that can be taken in response to increasing demand for computing courses and programs, and recommends strategies for institutions to consider in order to prepare for success in meeting their priorities in the middle and long term. Understanding this phenomenon and formulation of an effective response are crucial for the success of institutions, their students and faculty, and the nation.
1 In particular, the Computing Research Association (CRA) has led significant efforts to gather data, track enrollment trends, and stimulate awareness and consideration of this topic.
Since the emergence of the personal computer in the 1980s and the Internet in the 1990s, the field of computer science has seen an expansion of research areas and has been the driver of incredible economic growth, creating new industries and changing how society and businesses operate. Computation, from simulation and modeling to data mining, drives progress in many research areas, and has helped create new fields, such as computational biology, digital health, digital humanities, and more. Several key areas are generally viewed as having
direct impact on most other aspects of contemporary society and businesses: data science and machine learning, which enable gleaning of new knowledge from the vast quantities of digital information produced today, and cybersecurity, vital for keeping computing systems operational and protected from unwanted intrusion in the face of growing cyber threats.
Historically, computer science programs were formed as departments within some larger unit. Over time universities and other institutions of higher education have increasingly been creating colleges of computing, information schools, schools of information and computer science, schools of informatics, or other larger organizational structures. Today the nature and locations within an organization of computer science and related departments are varied, and depend on the size, history, and priorities of a given institution. Small colleges typically have computer science departments but no engineering programs, and some may combine computer science and mathematics, with courses for undergraduates only. Large universities may have both computer science and computer engineering programs, housed in colleges of arts and sciences or engineering, or in one of the newer, separate, computing-centric schools or colleges.2 Clarification of the terms related to the scope of computer science and related disciplines is provided in Box 1.2.
It is important to acknowledge that computing fields are grouped differently at different academic institutions, and that their boundaries have changed—and continue to change—over time. Tracking the temporal trends within a given field is thus a challenge, and the limited data available may be grouped in such a way as to impede such analysis.3 The discussions in this report aim to be inclusive, but in some cases are limited by availability and specificity of data. The committee explicitly states throughout which groups are included in any given analysis.
This study is concerned with undergraduate enrollments in computer science and related fields, both in the context of majors (students who declare as majors, and whose course of study leads to a degree in the field) and non-majors (students who take one or more computing courses but do not go on to receive a degree in the field). At a practical level, these categories cover enrollment in both computer science courses and degree programs. Information about the number of minors in computer science and related fields is also of interest; this is discussed where possible in the report, but this discussion is highly limited due to a lack of data.
Consistent with their charge, the committee sought out evidence from a range of sources. In addition to expert testimony and publications, several data sets were a source of significant input to the committee’s assessment; evidence
2 For a more detailed history of the computing disciplines, see The Joint Task Force for Computing Curricula (2005).
drawn from these sources is presented and discussed throughout the report. In the following, the source, content, and limitations of each data set are described as context for their discussion in subsequent chapters.
Integrated Postsecondary Education Data System (IPEDS)
Degree production at U.S. academic institutions is tracked over time by the U.S. Department of Education via the Integrated Postsecondary Education Data System (IPEDS), including those institutions in all U.S. states and territories and the District of Columbia. Degree completions and other information are reported by all institutions that receive funding under Title IV of the Higher Education
Act of 1965, as amended, with institutions self-reporting information that they classify according to the IPEDS taxonomy via the IPEDS Completions Survey (years available: 1966-2015) in general, and the IPEDS Completions Survey by Race (years available: 1977-2015) when considering race/ethnicity demographics of degree completions. This is the most comprehensive data set about U.S. postsecondary degree completion available, and it may be queried via the Web-based Computer-Assisted Science Policy Analysis and Research System (WebCASPAR) database maintained by the National Science Foundation’s (NSF’s) National Center for Science and Engineering Statistics.4 Two distinct data sets are available
for both surveys, one from NSF and one from the National Center for Education Statistics (NCES); the NCES data set is the source of data presented in this report.
The IPEDS tracks these data using a range of classifiers, such as institution type, and level and discipline of degree conferred. These classifiers have been periodically updated and adjusted to reflect changes in the IPEDS taxonomy as academic trends change. Academic disciplines are identified via its Classification of Instructional Program (CIP) codes, identified with varying specificity by six-digit (for the most specific classification of discipline) or four-digit decimal codes or discipline names, or a more general (termed “detailed”) classifier (e.g., “Computer Science”).
The detailed classifier of “Computer Science” includes all CIPs of the general form 11.XXXX, referred to as “Computer and Information Sciences and Support Services” (referred to in this report as CIS for short) in the “6-digit Classification of Instructional Program” categories (NCES, 2010a). This classifier spans computer science, information science/studies, and information technology, as well as other specific fields that may be more relevant to certificates or associate’s degrees.
11) COMPUTER AND INFORMATION SCIENCES AND SUPPORT SERVICES.
11.01) Computer and Information Sciences, General.
11.0101) Computer and Information Sciences, General.
11.0102) Artificial Intelligence.
11.0103) Information Technology.
11.0199) Computer and Information Sciences, Other.
11.02) Computer Programming.
11.0201) Computer Programming/Programmer, General.
11.0202) Computer Programming, Specific Applications.
11.0203) Computer Programming, Vendor/Product Certification.
11.0299) Computer Programming, Other.
11.03) Data Processing.
11.0301) Data Processing and Data Processing Technology/Technician.
11.04) Information Science/Studies.
11.0401) Information Science/Studies.
11.05) Computer Systems Analysis.
11.0501) Computer Systems Analysis/Analyst.
11.06) Data Entry/Microcomputer Applications.
11.0601) Data Entry/Microcomputer Applications, General.
11.0602) Word Processing.
11.0699) Data Entry/Microcomputer Applications, Other.
11.07) Computer Science.
11.0701) Computer Science.
11.08) Computer Software and Media Applications.
11.0801) Web Page, Digital/Multimedia and Information Resources Design.
11.0802) Data Modeling/Warehousing and Database Administration.
11.0803) Computer Graphics.
11.0804) Modeling, Virtual Environments and Simulation.
11.0899) Computer Software and Media Applications, Other.
11.09) Computer Systems Networking and Telecommunications.
11.0901) Computer Systems Networking and Telecommunications.
11.10) Computer/Information Technology Administration and Management.
11.1001) Network and System Administration/Administrator.
11.1002) System, Networking, and LAN/WAN Management/Manager.
11.1003) Computer and Information Systems Security/Information Assurance.
11.1004) Web/Multimedia Management and Webmaster.
11.1005) Information Technology Project Management.
11.1006) Computer Support Specialist.
11.1099) Computer/Information Technology Services Administration and Management, Other.
11.99) Computer and Information Sciences and Support Services, Other.
11.9999) Computer and Information Sciences and Support Services, Other.
It is important to note that the CIP 11.x series includes several fields that are not considered traditional CS programs but are more likely to be considered “information technology” or “computer or information support services.” However, these fields comprise a relatively small fraction of the bachelor’s degrees awarded at nonprofit institutions in the CIS category. For example, no bachelor’s degrees have ever been reported for “11.0203) Computer Programming, Vendor/Product Certification,” or any of the fields under “11.06) Data Entry/Microcomputer Applications.” However, the six-digit classifiers were introduced in 1987 and updated in 2000 and 2010; and there is no way to distinguish between all of the subfields over the whole time frame of 1966-2015.
It is also worth noting that, because a significant number of these degrees are classified as “11.0101) Computer and Information Sciences, General,” it is also
not practical to separate CS from information science completely—indeed, some institutions themselves may not treat them separately.
Given these factors, along with uncertainty in how institutions may self-identify the CIPs for their programs, the general variation in the nature of computing programs in the first place, discontinuities in the classification and reporting of subfields over time, and the difficulty in comparing cross sections of this category to national labor statistics data, the committee has chosen to present the whole 11.x series (CIS) in most of its analyses.
Other analyses of CS degree trends have examined a narrower set of classifiers; in particular, the Computing Research Association (CRA) has chosen to analyze only those degrees designated “11.0101) Computer and Information Sciences, General” and “11.0701) Computer Science,” in order to focus on core or traditional CS degrees in recent years (Camp et al., 2017a). There are several instances in this report where analyses regarding recent trends would be affected by using this narrower set of CIPs; these are noted and commented upon as the committee found appropriate.
Prior to 1987 computer engineering (CE) degrees were counted under other categories. CE degree production is tracked beginning in 1987, when it emerged as a unique classifier, as the 14.09 series. While some CS degrees may be categorized as CE (and vice versa), and CE is an important area of computing, the committee chose to keep it separate from CS to enable examination of time series trends.
IPEDS data are central to the discussions in Chapters 2 and 5. The specific data presented therein may be reproduced by an interested reader via the WebCASPAR system by querying the database with the same classification parameters used by the committee, as indicated for each figure in Appendix E.
Computing Research Association Survey Data
CRA Taulbee Survey
The CRA Taulbee Survey is administered annually to all North American Ph.D.-granting institutions that subscribe to the CRA. This survey collects computing departments’ self-reported bachelor’s, master’s, and Ph.D. degree trends. This annual survey has been a “principal source of information on the enrollment, production, and employment of Ph.D.s” in computing since it was first administered in 1974 (CRA, 2015). An adaptation of this survey has also been sent to institutions whose CS units do not grant Ph.D.s (known as non-doctoral colleges) since 2012 via the ACM, as a supplement to the Taulbee Survey.
CRA and NDC Enrollment Surveys
In the face of increasing CS bachelor’s degree production and major enrollment, in 2016 the CRA sent out a supplemental Enrollment Survey to “units” (programs, departments, divisions, schools, or colleges) in the Taulbee and NDC groups responsible for serving bachelor’s-level majors in computer science.5 This survey collected supplemental information about undergraduate major enrollments and course enrollments. The results were recently published online by CRA in a report titled “Generation CS” (Camp et al., 2017a, 2017b; CRA, 2017).
The CRA Enrollment Survey asked responders for the numbers of students (separated by majors and non-majors) in four categories of courses representing different points in computer science education: an introductory course for non-majors, an introductory course for majors, a mid-level course, and an upper-level course. Results were obtained for three years: 2005, 2010, and 2015.6 Overall, 134 of 190 doctoral-granting institutions surveyed responded, 45 of which reported course enrollment numbers (~24 percent response rate); 93 of the 706 non-doctoral institutions surveyed responded, 20 of which provided course enrollment numbers (~2.8 percent response rate). These are the best quantitative data available about current undergraduate computing course and major enrollments at U.S. institutions of higher education; nonetheless, they are not comprehensive, and, because responses were voluntary, they may reflect self-selection bias, so in general should be interpreted with caution.
The Freshman Survey of the Cooperative Institutional Research Program
The Freshman Survey of the Cooperative Institutional Research Program, (currently administered by the Higher Education Research Institute) has been administered to first-year, full-time students at a national sampling of 4-year colleges and universities since 1966, and covers a wide range of topics, including intended major. This survey is administered to freshmen “during registration, freshman orientation, or the first few weeks of classes.” The data are extensive, and designed to reflect the profiles of all new full-time students at 4-year colleges and universities nationwide. The results for fall 2016 reflect the responses of 137,456 students at 184 colleges and universities, weighted to reflect profiles of all new full-time students nationwide (Eagan et al., 2016). Statistics on student intent to major in computer science obtained from this survey between 1971 and 2015 are presented and discussed in Chapters 3 and 5. However, we caution that the uncertainties in these data are not easily quantified.
5 Units responsible for degrees in other areas of computing—that is, information science and computer engineering—were not included.
6 The CRA Enrollment Survey surveyed only computer science units, rather than computer engineering or information science. “Non-majors” here refers to students not currently enrolled in computer science degree programs (though it is possible that they may later choose to do so, and become majors).
Student Academic Data from the Consortium for Undergraduate STEM Success (CUSTEMS)
The Consortium for Undergraduate STEM Success (CUSTEMS) is a growing coalition of institutions for which enrollment data have been collected on a voluntary basis beginning in 2008 in order to track retention of women and underrepresented minorities in STEM degree programs.7 The CUSTEMS data set includes course enrollment, grades, and admissions records for students at participating institutions. Upon commission from the committee, the CUSTEMS team provided information about enrollment in CS courses at eight historically black colleges and universities, five liberal arts colleges, and one large public research university—all of the institutions for which the team had student-level data for the years of 2009 through 2014. The identities of the institutions remain confidential for privacy protection. Additional information about this data set is available in Appendix F.
The Current Population Survey (CPS)
The Current Population Survey (CPS) is administered jointly by the U.S. Census Bureau and the Bureau of Labor Statistics (BLS) on a monthly basis to a sample of 60,000 occupied U.S. households. The survey results are the basis for monthly (and annualized) statistics about the labor force, including occupational field, educational attainment level, and wages. The actual survey is conducted via in-person or over-the-phone interviews, with Census Bureau employees recording responses in a computerized system. Occupations are classified according to the BLS Standard Occupational Classifications (SOCs), which can be linked via a crosswalk to the taxonomies for the American Community Survey (ACS) and to the IPEDS CIPs.8
This report includes discussion of CPS data provided to the committee in a white paper written by economist Jennifer Hunt of Rutgers University, which is discussed in Chapter 4 of this report and included in Appendix D. Specifically, median wages and share of all U.S. employment are provided for computer and mathematical occupations (2010 SOCs 15-0000/2010 Census codes 1000-1240), engineering and architecture occupations (2010 SOCs 17-0000/2010 Census codes 1300-1560), and science occupations (2010 SOCs 19-0000/2010 Census codes 1600-1965). While the category of “Computer and Mathematical Occupations” is broader than those grounded in computer science skills, it does include most CS occupations, and is helpful for illustrating relevant trends.
The American Community Survey (ACS)
The American Community Survey (ACS) is administered by the U.S. Census Bureau to approximately one in 38 U.S. households each year.9 The responses of this survey are the basis for national statistics on topics such as U.S. jobs, wages, occupations, and educational attainment of the population. Individuals write in information about their occupation, which is coded according to a predefined taxonomy (the Census code) that is periodically updated, and which maps to the Standard Occupational Classifications used in the CPS. In Chapter 4, ACS data about the field of bachelor’s degree held by workers in computer occupations and the industries employing CS bachelor’s degree holders are discussed.
The following report presents available evidence about computing degree production, CS course enrollment trends and impacts, drivers of course and program enrollments, and diversity in computing. It outlines key findings, a spectrum of possible strategies for institutions experiencing CS enrollment increases, and general recommendations. Specifically, the report is organized as follows:
- Chapter 2 discusses historical and recent trends in computing degree production, and past actions taken by computing departments in response to growing enrollments.
- Chapter 3 examines recent trends in enrollment in computer science programs and available data about the impact of increased enrollments at institutions experiencing growth, as well as actions institutions are taking or considering by way of response.
- Chapter 4 explores drivers of the recent increase in demand for computing, including the labor market for computing and the changing landscape of computing in academia and today’s society.
- Chapter 5 assesses diversity in computing in recent years, and the relationship between enrollment growth and student diversity, including impacts of actions taken by faculty, departments, and institutions in response to increasing demand.
- Chapter 6 explores institutional needs and priorities, and possible actions to take in response to increasing enrollments.
- Chapter 7 presents recommendations for responding to the current enrollment boom and planning for the future.
This page intentionally left blank.