Interest in CS as a Major Drops Among Incoming Freshmen

Computing Research News ^ | 4/23/05 | Jay Vegso

[anymouse]

Survey results from the Higher Education Research Institute at the University of California at Los Angeles (HERI/UCLA) show that the popularity of computer science (CS) as a major among incoming freshmen has dropped significantly in the past four years. Alarmingly, the proportion of women who thought that they might major in CS has fallen to levels unseen since the early 1970s.

The percentage of incoming undergraduates indicating that they would major in CS declined by over 60 percent between the Fall of 2000 and 2004, and is now 70 percent lower than its peak in the early 1980s (Figure 1).

Freshmen interest levels at any given point have been an accurate predictor of trends in the number of degrees granted four to five years later. It therefore seems likely that there will be a sharp decline in the number of bachelor's degrees granted in CS in the coming decade. Results from CRA's Taulbee Survey of Ph.D.-granting CS departments reinforce this: the number of newly declared CS majors has declined for the past four years and is now 39 percent lower than in the Fall of 2000. Enrollments have declined 7 percent in each of the past two years (see www.cra.org/info/taulbee/bachelors).

Figure 2 provides a sense of changing interests among incoming freshmen. The majors included within the groupings can be found below. 

The upcoming drop in CS degree production will highlight the field's inability to appeal to incoming female undergraduates. Overall, interest in CS among women fell 80 percent between 1998 and 2004, and 93 percent since its peak in 1982.

Although newly-enrolled women have always been less likely than men to indicate CS as their probable major, the gap between them remained relatively narrow through 1980 (Figure 1). During the surge and drop in interest that occurred in the 1980s, however, the difference between men and women more than doubled. While their interest levels continued to parallel each other, it was at this time that CS appears to have lost its ability to attract incoming undergraduate women. During the second surge of interest in CS that occurred in the mid- to late 1990s, women's interest in the field did not grow at the same rate as men's. As a result, the gap between men and women who thought that they would major in CS tripled between the early and late 1990s. Although the difference might appear to have narrowed in recent years, this is because the percentage of women interested in CS was low to begin with, whereas men's interest levels have had room to fall.

Unsurprisingly, freshmen women's dwindling interest in CS has affected degree production trends (Figure 3). Unlike most other fields, which have seen women's representation increase over time, the portion of CS degrees granted to women fell in the late 1980s and has yet to return above 30 percent. With a fall in degree production looming, it is difficult to see how CS can match expected future demand for IT workers without raising women's participation at the undergraduate level.

---

Sources and further information:

HERI/UCLA's "CIRP Freshman Survey" is an annual survey of the characteristics of students attending colleges and universities as first-time, full-time freshmen: www.gseis.ucla.edu/heri/freshman.html.

National Science Foundation data on degree production are available at www.nsf.gov/statistics/ and on CRA's website at www.cra.org/info/education/us/

Fields included in Figure 2's groupings:

(Computer Science is categorized by HERI/UCLA in a group called 'Other', which was not included in this article).

Arts and humanities

Art, fine and applied
English (language and literature)
History
Journalism
Language and literature (except English)
Music
Philosophy
Theater or drama
Speech
Theology or religion
Other arts and humanities

Biological science

Biology (general)
Biochemistry or biophysics
Botany
Environmental science
Marine (life) science
Microbiology or bacteriology
Zoology
Other biological science

Business

Accounting
Business administration (general)
Finance
International Business
Marketing
Management
Secretarial studies
Other business

Education

Business education
Elementary education
Music or art education
Physical education or recreation
Secondary education
Special education
Other education

Engineering

Aeronautical or astronautical engineering
Civil engineering
Chemical engineering
Computer engineering
Electrical or electronic engineering
Industrial engineering
Mechanical engineering
Other engineering

Physical science

Astronomy
Atmospheric science (including Meteorology)
Chemistry
Earth science
Marine science
Math
Physics
Statistics
Other physical science

Professional

Architecture or urban planning
Home economics
Health technology (medical, dental, laboratory)
Library or archival science
Medicine, dental, veterinarian
Nursing
Pharmacy
Therapy (occupational, physical, speech)
Other professional

Social Science

Anthropology
Economics
Ethnic studies
Geography
Political science (gov't, int'l relations)
Psychology
Social work
Sociology
Women's studies
Other social science

Technical

Building trades
Data processing or computer programming
Drafting or design
Electronics
Mechanics
Other technical

[VeniVidiVici]

Testing can be more fun than pounding out code, or it can rival working on the dreariest assembly line.

Good luck to her finding the inbetween.

[Myrddin]

I have a friend at UCSD who was the resident "principal programmer" for 4.3BSD around 1986. His doctoral thesis resulted in a patent for the UC system. The topic was an algorithm to eliminate write delays on a RAID array. He was one of the first people in the world to do anything that could be reasonably called RAID. Subsequently, he designed the disk subsystems that are the core of the Walmart database systems. They turn over 95% of over 12 GB of data every day. He also has an BSEE and MSEE prior to the PhD in CS. There really is some very good stuff going on...if you look carefully.

[Myrddin]

... the best OS out there (LINUX) is basically the same danged OS that was banged out in berkeley 20 years ago (UNIX BSD).

BSD started life with a distribution tape of System 7 UNIX from Western Electric. Microsoft created Xenix on the 68000 (TRS16A) from the same tape. I still have a functioning TRS16 with Xenix in my basement. I used it in 1983 to bootstrap 10 UNISYS 1100 mainframes running UNIX*1100 from Bellcore. I had to fix some of the segment swapping code in the kernel to accomodate more users. I also had to fix the tty interfaces to the underlying OS1100 GCS subsystem. After 25 years as a UNIX professional, I still prefer that OS and the numerous imitators. QNX is my favorite embedded solution today.

[Myrddin]

"You know most of the greats on the pure research side did not have CS degrees at all - Knuth, Kay, McCarthy. A good many of them ended up with Turing prizes."

Actually its a fair observation. Mike Karels (chief principal programmer for 4.3BSD at UC Berkeley) has his degree in Microbiology from Notre Dame. I started life as a Molecular Biologist in 1976. Computer science and electrical engineering (all self taught) have served me well for almost 30 years. My colleague in Mclean, VA has a masters in Math from Stanford and a PhD from Yale. He writes some pretty amazing digital signal processing algorithms. Another colleague is a medical doctor. He excels at Oracle database design. He teaches the subject at a local college when not performing top quality work on contract at my office.

In general, a CS grad arrives for work with just a passing familiarity with a broad range of CS topics. Most lack experience working on team projects. What passes for a "big" project in a classroom is generally a trivial exercise for a sharp CS professional. In the process of hiring, I treat a CS degree as a sign of aptitude for the subject. The field moves so fast that you must constantly read the latest books to be current enough to be marketable.

[dr_lew]

It might have been more meaningful 20 years ago, but now a lot of those problems are solved.

Can I have the hardcopy on that please, HAL ?

[Myrddin]

One of the things I like about my job is that I program for micros where size and speed actually matter. A typical machine I program for has less than a millionth the RAM of a typical desktop machine, and runs at about a thousandth of the speed.

My current tasks are a mix of embedded Linux controlling a CAN network, GPS, 802.11b network, digital signal processing, mesh networking, ZigBee sensors, PIC microcontrollers and new actuators to couple/uncouple a train, control angle cocks and set/release handbrakes. The complete set of Linux OS plus applications has a disk footprint around 15 MB. CAN controllers have around 40 Kbytes of PIC code. It's takes a breadth of experience in embedded programming, datacomm, signal processing and electrical engineering at component level.

I've driven the PIC18F6585 down to 4 MHz clock speed with a 125 KHz CAN bus. The PIC plus transceiver draw is down to 14.5 mA. The factory configuration at 25 MHz was drawing 62.5 mA. Low power draw is critical for my project as it depends on power generated from a Timken bearing generator that provides 20 watts at 15 MPH to recharge the gel cell batteries on the freight rail car.

[glorgau]

Well, it isn't really designed at all. It happens by accident. Yep, most software organizations - and that includes most of Microsoft and Apple - are Capability Maturity Model level 1 organizations. They can repeat the feat, but can't really say how they did it.

[packrat35]

Maybe they don't want to move to India to work.

[glorgau]

There really is some very good stuff going on...if you look carefully.

This is where the US still leads the world - in operational management. Culturally, other countries just don't get it (yet).

[horse_doc]

I got my CS degree in 1985. I lived through multiple cycles of the old "we aren't producing enough CS majors" thing. When more students do major in CS, the industry just finds more ways to turn the screws on the expanded labor pool.

When that pool shrinks, a few years later, then they wring their hands again. Repeat ad nauseum.

I'm delighted that I ditched the industry forever in '98. And I would NEVER let my children major in CS.

[the gillman@blacklagoon.com]

Don't worry. There is a huge stockpile of former CS engineers working at Home depot in case we need to organize a draft.

[Starwolf]

It a heresy but I am not convinced that CMM and CMMI are all they are cracked up to be.

In the original version of Peopleware, Tom Yourdon talked about how in the software we are always thinking that the next new fad will save us from ourselves. Think he called it the Laetrile Effect or some such. I've been at it long enough to see several cycles of it.

The Gov requires CMM Level 3 or better from its contractors these days, but no commercial shop does the dance required to get the ratings. Note that I am in a CMMI & CM Level 5 organization.

[Starwolf]

Ahh. realtime. Where the rubber meets the road. My favorite area to work in. Takes real skill and knocks out most of the cross trained arts & parties majors.

[Lizavetta]

A good student will need no more than 5 years to get a PhD

But that would mean the little darlings would have to take classes at 7 a.m. or in the evenings or they'd have to go during the summer. That's so unfaaaaaiiiiiirrrrrr.

[bert]

.....The good news is that enrollment in the CS dept. at Banglor Tech ....

The bad news is that Bangalor Tech is a high school

[kdot]

Amen to that.

Besides, our so-called "Computer Programmers" today have alphabet-soup resumes and a wonderful grasp of software development packages. Sure they can chunk out code. But they couldn't begin to discuss kernal mode operations, distributed lock management, device drivers or any multilayer architecture. No, none of that matters because they just have to use the pretty GUIs to establish their locking strategy - or click-here to set up the network.

The reason CS is important is because we still need programmers who can write the next generation device drivers for the new-fangled drives and periperal devices. We still need our operating systems to evolve so the shiny development environments can support high-pay project manager jobs.

Most software developers today are users of robust GUI environments. Great jobs - gotta have them. But real CS programmers know words like MUTEX, QIO, CMKRNL, context switching, paging, swapping, etc.

I know, I'm a dinosaur.

[WOSG]

Yes, I know the history, having started in the VAX/UNIX era. The Red Hat distro LINUX I have on my box right now has evolved from that heritage, and, sure, if it aint broke dont fix it ... but my point was that the pace of evolution is SW is so much slower than hardware. And I speak as a UNIX/LINUX fan and advocate. It could be worse, windows could have take n over the world. Thank God for Linus Torvalds. :-)

In retrospect, that may be a bit unfair towards SW innovation, as the Athlon is still derived from x86 architecture, so even thought the athlon is 10,000 times more powerful than the 80286, the ISA system engineering has been more incremental.

There has been a lot of innovation in CS, but it seems the processes of standardization is more powerful than innovation... this I think is a direct consequence of Moore's law, which has the side-effect of being so powerful that innovations are irrelevent. e.g. remember 'lisp machines'? Great idea, except that they became slower than PCs and standard-product-based Sun workstations pretty quick. And look at supercomputers today: Take thousands of Opterons and slap 'em together. Someone engineering a careful optimized vector CPU will get crushed (notethe many failed supercomputer/paralell-computer firms).

[WOSG]

"In general, a CS grad arrives for work with just a passing familiarity with a broad range of CS topics. "

Ah, but that shows how deep the topic really is!

A software programmer doesnt need 90% of what CS is about (from algorithms to computer engineering) to do his work.
Yet I am 15 years post PhD and am learning every day, because the field I am working in (EDA/CAD) is evolving.

"The field moves so fast that you must constantly read the latest books to be current enough to be marketable."

Agreed... and that is something few other domains can claim.

[Cvengr]

But that's because you came into the field back when people thought Turing Machines were BMWs ;^)

[Smogger]

Sometimes I question the value of a strict CS degree.

Practically worthless, none of the programmers I know have them. Few jobs require them. Get a degree in business if you are a good programmer, that way you can always start your own after working for someone else for awhile.