I apologize to my very few readers for my lengthy absence. (I’ve been busy, and my cats ate my last post just after I finished writing it.It was my best post ever.) Realizing that this will be a post of limited interest to the general public, I have included the usual gratuitous pictures of dishes from the 60s. Please read the text in between them, folks.
And there will be a quiz at the end. Winners will be acknowledged.
CAJOLING AND CARESSING THE DATA
As part of a real MPH (the concentrations where you actually learn to do something, such as epidemiology, biostatistics, environmental health, health law, etc, as opposed to the amorphous field of international health) most students are required to take a semester of biostatistical computing. This should probably be at least 2 semesters, especially for epidemiology concentrators, but that’s a topic for another time. At the august Boston University School of Public Health, that means a semester of SAS.
SAS (formerly Statistical Analysis Systems), is the software produced by the SAS Institute , based in North Carolina. SAS software, currently in version 9.2. It is used for not only statistical analysis, but data mining, IT management, business intelligence, etc. They are clearly doing something right. For two years in a row, the SAS Institute has been named Fortune magazine’s best company to work for, and their software could easily be considered the industry standard in its particular area.
Fine and dandy. There is a hitch of course, and the hitch is that SAS licenses are expensive. Really expensive. For those of us in Public Health, they are Ferrari and Audemars-Piguet expensive. I called to see what it would cost me to get a 1-year license for a single user to use SAS (are you sitting down?)
For the projects that I work on, that’s way too expensive . Last year I did some work on Ehrlichia in Peru, where the whole budget, including paying the staff, was less than $3000. I just completed working on a data set regarding the behavior of cats in shelters for a non-profit organization. Using SAS on this would have been like chartering a luxury yacht to go commercial fishing–it just can’t pay off.
Clearly, SAS is worth it for some users, but I can’t speak to that. What I do know is that if the biostatisticians continue to teach SAS in public health schools, they will be instructing their students in a technology that could easily be out of their reach.
WHAT ARE THE ALTERNATIVES?
I just finished crunching data for the ASPCA. I used a free 30-day trail of Minitab 16 that I downloaded from the internets. It was more than sufficient for my project. The data set, though, was rather small, consisting of 40 or so variables on 253 cats, taken at 5 different points in time. The downside was I had to take out a mothballed PC, and I am a Mac person. (BTW, both SAS and Minitab used to make versions that could be used on a Mac. I really like using my Mac. It is far more stable than any PC I have ever used. ) One of the other things that I like about Minitab is that while it uses a menu-driven, point-and-click user-friendly interface, one can enable the command editor and the code will be recorded in the session window.
What’s the best endorsement for Minitab? The PhD in statistics who helped me, Al Kabong (not his real name), prefers Minitab over the other statistical packages available to him. He admits that for monstrous data sets he is going to need SAS, but he is an emeritus prof and has the resources of a large research university behind him.
Is the system rigged against small projects?
Now for the quiz: Who are the sassy chickadees in the pictures?
Next issue: JMP , R, and the answers to the quiz