[Macworld]

Post your comments for SPSS 16.0 here

[Ejames]

It should be noted by all potential buyers that SPSS has treated its academic Mac users VERY poorly. After collecting licensing fees they failed to update programs to work on Intel machines, saying that it would do it in a year's time! This left many users and labs with Intel Macs with useless software for a year. SPSS refused to even talk about a refund or adjustment for the unusable software. They ought to make good for a years' worth of product. It's a shame they are the only game in town. I would like to see the open source folks put these guy's heels to the fire.

[folklore]

Ejames said:

It's a shame they are the only game in town.

You're right that SPSS's Mac support is abysmal - I once asked a trainer (employed by SPSS) about the Mac version, and her answer was, "We have a Mac version?" The Mac version has historically lagged behind the PC version by two versions and about two years. Literally. Their version 13 for the Mac was released around the same time version 15 was released for the PC.

You're wrong to say they're the only game in town, though.

Stata is available for the Mac (and Windows, and Linux/Unix).

And last I checked, Stata could perform some analyses that SPSS cannot, though the last time I checked was a few versions ago. Stata is certainly worth checking out. I don't use it for the same reason I don't use MS Office alternatives - all my colleagues use SPSS.

[tmicceri]

Yes, I have been Using SPSS since 1980 myself on IBM Mainframes also using JCL (Job Control Language) - those using DEC machines didn't need to worry about that.

And Yes, SPSS's support for the Macintosh Community has been poor. Their licensing methods are also a real pain.

However, SPSS 16 has become a really useful tool of Choice for a variety of reasons, in addition to the fact that I think it is pretty elegant for the unsophisticated user (not in the same ballpark as StatView used to be, but in many ways modeled on that application):

1 - It handles large datasets very well (I frequently work with over a million cases with hundreds of variables and have not seen a hiccup - admittedly in a PC environment up till now - haven't yet tested it on the Mac).

2 - It can convert just about any type of File to any other type of File. I therefore use it for conversions. Just before the holidays, I converted a massive database which I created using SAS into SPSS, .dbf and .xls formats. SPSS has dedicated itself to be the cross platform/cross format converter.

3 - SPSS has some statistical methods that are difficult to find elsewhere unless you use a dedicated application. Among those are (1) reliability and (2) multiple response (when respondents can give more than one answer to the same question - common on Surveys and Job Analyses).

4 - SPSS has what I consider to be far and away the best programming environment for Multiple Regression and Logistic Regression.

Also, SPSS has developed what I consider to be far and away the best Text Analysis for Thick Response (http://www.spss.com/textanalysis_surveys/).

Unfortunately, in the Mac Community, what we have seen through the years is that the Big Guns (SPSS and SAS) have gobbled up the smaller, localized software firms. The two that bug me the most are Abacus Concepts (Statview, SuperAnova, MacSpin) and DeltaGraph Pro (still available through RedRock Software (http://www.redrocksw.com/deltagraph/mac/).


Of course, if you are a programmer (coder), then R is a choice. R is the Open Source (Free) version of Bell Labs S system, unfortunately, it is equally difficult to use. Source:(http://www.gnu.org/) - a couple of other sites for

Free Statistical Software are: http://www.statsci.org/free.html, and http://StatPages.org/. There is also links to freeware and a couple of Freeware Applications I have produced on my Statistical and Other Myths web page: http://www.freewebs.com/tedstats

So, I think that SPSS is an excellent company and tool (particularly their Student Versions - the Graduate Version is pretty much the complete SPSS application, with basically no data size limits).

However, it is true that SPSS has failed to properly support we Mac users in the past (of course, the vast majority of statistics users use PCs).

[edwardayres]

I agree that SPSS treats it academic customers poorly. You must buy the program then abandon it after graduation; their license only covers its use while you're in school. When someone graduates the last thing they have is enough cash to go out and re-buy the program for $1500. I stopped using it years ago because of these things.

[Dan Frakes]

Ejames said:

It's a shame they are the only game in town.

As folklore noted, SPSS isn't the only game in town: Stata is an excellent package. Although I can't speak to the current versions of each, a few years ago I used both regularly. While SPSS's menu-driven interface was better than Stata's, I personally preferred Stata, especially when it came to writing programs -- the syntax was much easier to learn and was more versatile. The Stata user community is also a great asset, providing quality add-ons for various types of analysis. (Although, again, I haven't used SPSS 16 or the current version of Stata, so take these comments in that context.)

[lithgow]

I would like to use the new version (16.0) and write a great review.

However it has been "Stuck on a ship" in Singapore for the past 2 months (I am in Oz). I am never paying for software up front again. It is compulsory to use SPSS for my Psychology courses ? how they expect you to do this without the software I will never know.

At least the Grad pack is only about $200AUD...

[BRussell33]

Can anyone tell me if the basic version of SPSS 16 does repeated-measures ANOVAs, or mixed factorial designs (with both repeated measures factors and between factors)? Thanks

[barrett]

I've used SPSS on the PC and also Quantum (more powerful data and tab s/w also from SPSS) on a unix box for nearly 10 years. In fact, SPSS 15 is one of the reasons that I still use Parallels. I am trying to make the upgrade to 16 because I would rather pay their upgrade fee rather than another year of their license fee.

Yes, as a software company I would have to agree that SPSS leaves a lot to be desired. SPSS has been notorious in the past for over promising. They can do a pretty incredible demo for you on-site. Don't expect anything that they showed you to be available anytime soon, if at all, though. I found them to be masters of smoke and mirrors in that regard.

They have also been really bad about releasing software to you when it wasn't production ready. We spent 2 or 3 years as part of their development cycle. They would release a version of something, and we would find critical bugs or shortcomings. They treated us as if we were their beta testers, but we still got to pay full price for their goods and services.

Given the choice I would not use SPSS products, but as I work in marketing research I am left without a choice.

[acock]

As SPSS has moved away from academic statistics to more business applications it has failed to keep up with statistical developments. Mac people should look at www.stata.com. This package puts entire datasets in memory so it runs vastly faster than SPSS. It has professional graphics far beyond the gee-wiz marketing graphics in SPSS. Its statistical capabilities vastly exceed SPSS. Its price includes the entire package unlike SPSS where there is a Base statistics, then regression statistics, then advanced statistics, etc. Universities with an agreement with Stata get it at a tiny fraction of the price of SPSS. Stata is extensible by users and there are 1,000s of specialized commands written by end users that are installable on the fly.

Here is one example. SPSS charges extra for a missing value imputation module that does single imputation using EM. An article in the American Statistician pointed out that it was incorrectly implemented and to my knowledge, SPSS hasn't made the corrections for a few years. A Stata user wrote a command called ice that does multiple imputation instead of single imputation (better than SAS's micombine) and uses an appropriate estimator (regression, logistic regression, multinomial logistic regression).

Stata has a menu system like SPSS, but its command structure is so much simpler than SPSS that regular users skip the menus except for graphic commands that can be complicated. For example, compare the following commands to what they would look like in SPSS
Regression
regress y x1 x2, beta
Logistic regression
logit y x1 x2, or
Poisson regression
poisson y x1 x2
Zero inflated Poisson regression
zip y x1 x2
Multinomial logistic regression
mlogit y x1 x2
Zero inflated negative binomial regression
nbeg y x1 x2

The reason for this simplicity is that Stata was developed for PCs and Macs and not for the time sharing mainframes of SPSS' hayday.

Alan Acock

[mvallance]

Thank you Alan for a great summary. I used SPSS when I was at a uni in Singapore and even though there was a site license, SPSS would not provide for Mac users..only Windows. I have not been happy with SPSS since I bought the previous version for my Intel macBook and found out it did not work -- Ok I should have checked but now being in Japan I had to order through my university and paid a steep "Japanese " price . Now I can't even upgrade and am being asked to pay the same amount for the new version which does work with Intel MacBooks. No thank you. So I was looking for an alternative when I read your post. Cheers and a Happy New Year to all Mac users :-)

Michael

[acock]

Michael,

Check out
http://www.ats.ucla....reports/number1editedFeb22007/ucla
ATSstattr11.1_0207.pdf

For a great comparison of Stata, SAS, and SPSS

[bntville]

In 1997, SPSS, INC. approached our Department with an offer of a 100-user Mac site license and 4-year maintenance deal with the promise of significant upgrades to the Mac SPSS v.6.2.1. The price tag was a little under $15K, which we considered to be a fair price for academic classroom usage. Within two months of this purchase, SPSS promptly announced that they would no longer support upgrades to the Macintosh platform, and we were essentially left to the flounder for almost 4 years with the woefully crash-prone v.6.2.1. During this period, this application was so troublesome that many of our instructors abandoned it in favor of other statistical packages (Statview, SuperANOVA). Our technical staff spent an inordinate amount of time dealing with these issues, generally to no avail, and SPSS Tech Support was woefully inept at resolving any of the numerous bugs and system incompatibilities. I did, however, get to listen to a joyful amount of Musak.
The long-overdue SPSS v.10 for Mac eventually alleviated some of our users' aversions to using SPSS in the classroom, despite the fact that the program continued to be highly crash-prone. SPSS then approached us with a request to renew our "maintenance" contract for 3 more years for an additional $15K. After considerable debate, we elected to do so. This entitled us to (eventually) receive SPSS v.11 for Mac which worked tolerably well for about a year.
At the end of this period SPSS once again requested $5K for another year of maintenance, which we paid. At this time we elected to upgrade our 100 workstations to the new Intel-based Macs. Shortly thereafter, SPSS promptly announced that not only were analyses executed with v.11 "unreliable" on the Intel Macs, but SPSS v. 13 would not even open on these computers. With absolutely no hesitation (or shame) SPSS, INC. announced that a solution to this fiasco would be a full year away, which left us with 100 unusable licenses. Add to this the fact that the Intel Macs would not run OS9 apps, we also no longer had the option to fall back on Statview or SuperANOVA. We managed to survive the year on a meager diet of R, some open-source apps, and a handful of Windows SPSS licenses that we, naturally, had to rent from our institution.
We did just receive SPSS v.16 and it does seem to actually open on an Intel Mac running OS 10.5, but we have not thoroughly tested it, and we remain skeptical. Naturally, SPSS, INC is now asking for another $7K for an additional year's "maintenance". And we may just have reached the end of our rope. We WILL be lookin closely at Stata and any other solutions that can free us of our dependency on SPSS, INC.

--bntville

[acock]

If considering Stata, check the GradPlan prices. These are available to any university that thinks (no promise) it will sell $2,000 a year site wide.
http://www.stata.com...u/gradplan.html
When considering these prices, remember that you buy Stata and it is not an annual license. The current version is Stata 10. I know people who are happily using Stata 8. There is no maintenance fee. Stata is updated at least monthly for the current version and this is automated on the internet.

In addition to the individual purchases, we have 30 concurrent users of Stata on a server. Anybody who has a university account (about 25,000 people at my university) has free use of these from any computer connected to the internet--so long as no more than 30 use them concurrently. I think the charge is something like $75.00 for each addition to the 30 user restriction.

There is a great support in the list serve at
Do you know about the independently operated Stata listserver? Hosted at the Harvard School of Public Health, Statalist is an email listserver where over 2,500 Stata users from experts to neophytes maintain a lively dialogue about all things statistical and Stata.

If you have an Internet email address, you can subscribe, and it?s absolutely free. You do not have to own Stata to join Statalist.

* To subscribe send an email to majordomo@hsphsun2.harvard.edu with subscribe statalist in the body of the message, not in the subject line. The subject line does not matter.

* After you subscribe, you will receive email as a member of the list.

You can join the discussions, or you can just listen.
To post to Statalist, send email to statalist@hsphsun2.harvard.edu.

Additionally, I've sent emails to Stata's own technical support and had Ph.D. Statisticians provide detailed answers where they even generated datasets to illustrate the solution.

Alan Acock