Elliott C. Back: Internet & Technology

If you read the news, you inevitable come across articles like Want a job? Get a computer science degree, which suggest that computer science “grads received an average of 2.3 job offers and had an average starting salary of more than $72,000 – the highest of any starting salary.” Surveys by NACE, such as Top-Paid Majors for 2009-10 Bachelor’s Degree Graduates, rank Computer Science as the fourth most lucrative undergraduate degree, after Petroleum Engineering, Chemical Engineering, and Mining & Mineral Engineering. With rising tuition costs, undergrads are feeling more pressure to graduate with majors that will help them repay their student loans and establish themselves.

I noticed that many universities post the results of their alumni employment surveys online, and gathered them into a spreadsheet. I will be looking at Cornell University, MIT, Stanford, UC Berkeley, CMU and UoA. I want to treat each university as a kind of asset, and categorize their volatility and average return. I imagine that the kind of university a prospective student, interested in maximizing his first-year earning potential, would choose would offer the most resilience to downturn, and the highest absolute salary.

Salary over Time per University

It’s clear from the above graph that the University’s ranking is a factor in first-year salary. The cluster of “top private schools” pulls away from the University of Arizona in earnings potential, and maintains an average of 23% more salary dollars in the last 12 years. That’s a significant spread.

Given the downturn during the tech crash of 2001, and the recent financial crisis, it’s worth asking which University will give you better returns than inflation on your salary (all of them), but without too much volatility (standard deviation of returns). As it turns out, there’s only one university in that sweet spot: Cornell University (disclosure: my alma mater).

While Stanford/MIT/UC Berkely have average salary growth of 4.5% per year, they also have an average standard deviation of 8.5% and took a whopping -12% hit in the downturn of 2001. Cornell got by with a -7% return, but has a much more respectable standard deviation of 5% and average salary growth of 4%. Carnegie Melon has a similar deviation in returns, but only with a 2% growth rate.

This is informal analysis; if you have any comments or questions, please drop a note on this post! One obvious shortcoming of this data is that it only records the starting salaries of undergrads, which isn’t a good indicator of total earnings potential for undergraduates of various universities. In the longer view, there may be no difference between undergraduate universities.

Appendix of Data:

• Inflation data from Consumer Price Index – All Urban Consumers – (CPI-U) – U.S. city average.
• Cornell University data from Annual Starting Salaries 1985 – 2006 and 2008 Employment
• MIT data from their salary survey (only three years, WTF!), and graduating student survey.
• Stanford has a pretty good index page 1997 to 2008.
• UC Berkeley’s What Can I Do With a Major In…?.
• CMU Post Graduation Survey Results from the School of Computer Science.
• University of Arizona’s Career Destinations Results page.

Data is based on mean starting salary. Data is linearly interpolated to form smooth curves where knots are missing. I was unable to find data for Columbia or Princeton, so if you have a source, please let me know and I will update my spreadsheet!

I came across PeopleSoft Hinders Review of Aid Applications, an article describing how my Alma Mater’s implementation of Peoplesoft is causing delays in processing financial aid applications, which contains an amazing quote:

12 days after classes started, about 750 students’ financial aid applications are still being processed due to complications from the implementation of PeopleSoft. PeopleSoft replaced JustTheFacts software and now manages students’ personal, academic, bursar and financial aid information.

“The PeopleSoft system is much more labor-intensive than our previous financial system,” stated Davis. “We estimate that it takes three to four times longer to review and process a financial aid application in PeopleSoft than it did in our former financial aid system.”

Well, “This can’t be that bad,” I thought and headed over to the student management system to check it out for myself. Then, I got hit by an ancient-looking ugly, non-functional GUI:

Clicking on any of the links that might interest me (grades, transcripts, etc) led me to the following “nice pages”:

The back/forward buttons don’t work

Whoah, where did this come from?

Another random error

Yep, Peoplesoft definitely sucks. And, I don’t blame Cornell for it–except for making the original bad decision to migrate from a working, if not archaic, system. Nay, I blame Peoplesoft (recently rebranded as Oracle Peoplesoft. Here’s some juicy quotes from other IT professionals:

• “It’s a horribly designed piece of crap. They don’t use referential integrity *and* they duplicate data all over the place in the database. Their UI is like something out of Windows 1.0 days.” – Joel on Software
• “It is the single worst example of web-based software I have ever seen. Ever.” – Jason
• “It’s web browsing in the 19th century. Lots of backing up and clunky navigation menus laden with far too much non-intuitive information.” – Dee-Rob
• “the syntax, which seemed arcane at best and totally unusable at worst” – John

A cute Facebook group, Cornell must be held accountable for Peoplesoft issues, and another Cornell story, New CoursEnroll Software Causes Distress, Difficulties, explains how the initial rollout was also fraught with difficulties:

Yeh said that although the system was marked by a number of problems, nearly 3,200 students out of the approximately 3,700 who needed to enroll were able to do so successfully. Course requests that were not approved before the system went down were entered into the system automatically.

[T]he preparation for the replacement began back in 1995 when Cornell administrators began watching how other schools used PeopleSoft. Cornell and company began to develop the new program together. Yeh did not know how expensive the program was to develop.

The moral of the story? Sometimes rolling your own software is better than buying and adapting. Especially for giant applications.

If you haven’t had a chance to read Why Black Nerds are Unpopular by David Adewumi, you should run over there right now. It gives an interesting cultural explanation for why the author believes we don’t see many African Americans in IT / Computer Science. It’s the inspiration for this post, a sort of state of the world of black diversity in IT. In his article, David writes how few of his black friends are “nerds:”

I would say, as a young black male, there is a strong inverse correlation between being a nerd and black, and being popular. I’ve seen many black friends who are fairly intelligent that were mediocre students in high school, and either failed out or were equally mediocre at the University level. Why? Popularity is, as Paul mentions, often times a choice of priorities — some sacrifice intelligence for popularity — and for blacks, this probably happens for 9 out of every 10.

I would go so far as to say that the lack of black nerds is probably a cause for major concern, but within the scope of this writing, possibly too large a problem to properly address, although certainly an interesting one.

After some Googling, I was able to find data from the National Science Foundation (NSF), Science and Engineering Degrees, by Race/Ethnicity of Recipients: 1995-2004, with information about degree recipients partitioned by self-identified race:

In 2004, 5,934 black students out of 57,405 total (10.33%) received undergraduate degrees in computer science. Overall, among all degrees, 4.84% of black students chose a degree in Computer Science as opposed to just 3.15% of white students. I don’t have enough personal or intellectual background to discuss these figures, but to my uninformed eye, they look quite promising. More blacks (by percentage) are choosing to study Computer Science than whites (our baseline majority in the US). And, while at 8.4% black undergraduate students feel underrepresented, the news indicates graduation rates are improving.

America needs to moving forward on providing excellent education to all Americans, not just the privileged majority. Perhaps our next President–who looks to be Barack Obama–will be tougher on education than his “no child left behind” predecessor and we’ll see these numbers get even better.