The other day I stumbled upon this graph, which shows the relative popularity of statistical softwares as a skill requirement in job postings. I was shocked that the top runners are SPSS and SAS, whose programming language is an ad-hoc, inconsistent mess that pales next to that of R. The definite lead of SAS over R is even more surprising considering that, whereas R is completely free, a single licence of SAS cost $8,700, plus a 28% annual fee ($2,436 / year).
Who in their right mind would reject a free, superior product and pay to suffer a clunky piece of software?
Judging by the graph, a lot of people indeed, and they do have working minds. Decades of SAS code is expensive to be converted into R, especially since the programming logic (not just the language) is different between them. Furthermore, unlike in academia, where breakthrough is richly rewarded while sloppiness at the edge is tolerable, industries place utmost value upon a stable and predictable performance. (You wouldn’t be comfortable to be sky high in an “experimental” airplane, would you?) Indeed, there are valid reasons besides the “cheap = bad” mentality to prefer SAS over R, even if we think of companies as a profit-maximizing entity.
What if we think of the firm as a hodgepodge of self-interested, ass-covering individuals instead? Then the cost of SAS also makes cynical sense — it is a great excuse. If the code goes wrong, the mid-level manager can righteously claim that he has ordered the most expensive and popular software in the business. Indeed, how could he have known that something would go awry? The fact that R is free suddenly becomes a negative. It leaves one vulnerable to incrimination, by oneself or others, that “you could have done more.”
This logic closely resembles what I have encountered during conversations about the free SAT database project, mentioned in an earlier post. Like SAS-vs-R, commercial SAT classes are popular, costly, while not being superior than self-study. But most data analysts are aware of the inferiority of SAS — they choose it anyway because of other reasons. Maybe parents stick with SAT classes because they are less informed?
So I informed them. Given how much they paid, I anticipated outraged denial and yet there was very little. They already sort of knew. When paying huge sums to SAT centers, they are really buying the relieving assurance that someone is looking over their kid, that there is someone to talk to just in case (though few actually avail themselves of this service.) Changing their mind will not be easy. While parents don’t suffer as much from the kind of inertia endemic in organization, nothing is more parental that the worry that “I could have done more.”
This realization upends my initial ideas about the project, which I envision to include an “awareness campaign” of sort about SAT classes. But awareness is already not in short supply. Parents’ expense on SAT is already an informed decision, one that aims to bring, besides the best things for their kid, the assurance that they have done the best for their kids.
With all honesty, seeing people making such decisions give me a slight impulse of frustration. $60 / hour just to have someone recite answer keys and watch over your grown-up kid? (That’s 6 times my pay as a writing consultant at Colgate, by the way.) If I were free to molt and cast my ideal universe, no one would do so.
But this world is the one that counts, and in it parents want reassurance. And thus we must provide, not question, for our goal is to help, not criticize. I am thinking that the SAT answer database cannot be a standalone product, floating in unmarked places run by unknown people. That does not sound reassuring, does it? Instead, it must be incorporated into a study group model, and a structured one at that, with student sorting, progress monitoring, and incentive scheme to punish slacking-off. Building credential is vital, and key to this is a physical and institutional presence (and some, I hate to admit, publicity stunts, such as showcasing test scores and achievements of the people involved).
At the moment, I am looking for student groups that would be able to implement this project. VA Club seems like a good start.