The Disappointing State of SPSS

I recently installed SPSS, what is generally considered the premier statistical program for social scientists, onto a colleague’s Intel-based MacBook Pro. Rather than install SPSS for the Mac, we opted for installing SPSS for Windows (version 15.01) on a Bootcamp partition on the MacBook Pro. One reason for this is because our campus has a site license for SPSS for Windows, but not one for SPSS for the Mac. But more importantly, SPSS for the Mac is not currently supported by SPSS on Intel-based Macs (which includes all Mac models released in the last year).

It seems rather odd that SPSS would post this rather confusing disclaimer about SPSS for the Mac on their website:

SPSS does not support the use of any existing version of SPSS for Mac OS X on the new Intel®-based Mac hardware, including SPSS 11.x or 13.0. The use of the Rosetta emulation software interferes with the numerical calculations in SPSS. We therefore are unable to support any version of SPSS on Intel-based Macintosh machines.

In other words, if you want to run SPSS today on a new Mac, you have to run SPSS for Windows using a bootloader (like Bootcamp) or virtualization software (like Parallels or VMWare). Running SPSS for Mac OS X under Mac OS X on a new Mac is not supported, at least not now. SPSS does state that they plan on releasing sometime during 2007 a version of their program that will run on Intel-based Macs. For now, however, the most viable option for Mac users who need to run SPSS on a new Mac is to run SPSS for Windows using Bootcamp (which technically is still a beta product).

It doesn’t take much to read between the lines here: SPSS doesn’t really care much about Mac users, and by extension, academics who use Macs. As someone who painfully remembers the period in the late 1990s when SPSS basically abandoned the Mac market, I can’t say I see this as a positive sign. But I’m not surprised. SPSS has never really impressed me as a company that cared much about the academic market, where many Mac users work. Apple has always been a company that takes pride in serving the educational market. SPSS, on the other hand, has consistently seemed to treat academic researchers as a small, and rather pesky, niche market.

It also doesn’t surprise me that our campus’s Academic Computing support specialist for SPSS doesn’t have a very high opinion of the product, regardless of which platform it is on. He points out that there have been quite a few well-documented bugs in SPSS software, and many have questioned the reliability of the software (see, for example, the article by McCullough in American Statistical Review). While the company does try to release regular patches to address those bugs, it doesn’t exactly contribute to user confidence in their product. After I installed SPSS for Windows on my colleague’s MacBook Pro, I got a couple of tersely worded e-mails from this so-called “support specialist” in Academic Computing, basically stating that he won’t support SPSS for Windows on a Mac. Period.

While I’m disappointed at his resistance to support this product, I can’t blame him for his attitude toward it. Over the years, SPSS has become a bloated piece of software driven more by the demands of applied researchers in business and industry than those doing “pure research” in academia. Just take a look at this “customer stories” page on their website. Only a small percentage of these “success stories” come from academics. And many, if not most of the academics that are listed on this page are from researchers in business and marketing schools, not from the more traditional “social sciences” that gave SPSS-the Statistical Package for the Social Sciences–its original name. Indeed, searching for the phrase “statistical package for the social sciences” on the SPSS site yields just a couple of brief mentions of this original name on their corporate history page.

Perhaps the reason SPSS hasn’t been paying much attention to the academic market in recent years is because there is a lot more money to be made in selling software for “data mining” and “predictive analytics” than in serving the needs of theory-driven research. It could also be because academic researchers can be tough customers, who expect a lot from their software, and typically demand a hefty educational discount (which SPSS begrudgingly provides, although their educational discounts aren’t nearly as deep as those provided by Microsoft, Adobe or Apple).

It would seem to me that if SPSS isn’t careful, more and more academics might opt for open-source software alternatives. Micah Altman maintains a good list of free (or cheap) alternatives on what he calls “The Impoverished Social Scientists Guide to Free Statistical Software and Resources.” It’s worth checking out. So is one of the most mature open-source statistical programs available today, Macanova, which has been around since 1987. Another program to watch is PSPP, which is one of the most “SPSS-like” open-source choices out there. And there seems to be a swelling fan base for the R statistical language, which received this glowing review in a direct comparison to three major commercial statistical packages (SPSS, SAS and Strata).

Many, if not most of the open-source statistical programs are Unix-based programs. This means they typically can run just fine on a Mac (since Mac OS X is, after all, Unix at its core). And while open-source software can and often does have bugs, at least the open-source community is honest about it. Bugs are addressed in hours rather than in months, and new versions of the most popular open-source programs are often released on a daily basis as “nightly builds.”

Of course, for the average academic researcher, trying to find just the right open-source tool for the job can be almost as frustrating as waiting for SPSS to release a bug-free version of their software. I really can’t blame academics who just want to get the data out the door and into publication as expeditiously as possible. And I really can’t blame academics who cut their statistical teeth on SPSS in grad school for wanting to stick with what they know. But I can blame SPSS for their increasingly evident insincerity in serving the needs of academic researchers.