Despite its hyper-modern Aqua look, SPSS has been with us since the days of punch-card computing, and this is reflected in its traditional product structure. There’s a £795 Base program, and then a series of add-on modules sold separately – Advanced Models, and Tables, Trends, Regression, Categories, Conjoint, and Missing Values Analysis. If you’re a social scientist dealing with field surveys, you might buy the Base program, Tables, and Missing Values Analysis (there are always plenty of non-respondents to surveys). If you’re an investor trying to predict what the price of soybeans will be next March, Base and Trends could be all you need. The modules integrate seamlessly into the main application, appearing as new commands under the Analyze menu when you enter the module authorization code. It’s wise to carefully consider which modules you need, since the whole all-modules package costs upward of £4,000. This makes it the most expensive statistics system available for the Mac. Although SPSS’s Mac user-base leans toward academics, this product is also a business-statistics powerhouse that can manipulate and analyze large, complex files. The first step in any business analysis is getting the data into usable form, so SPSS reads Microsoft Excel and other spreadsheet files directly, and other data files (MySQL, Microsoft Access, and so forth) through the ODBC (Open Database Connectivity) drivers provided on the SPSS installation CD. The data appears in SPSS’s spreadsheet-like Data Editor. But far superior to hand-coding with commands is the new Restructure Data Wizard, which can take a data table of arbitrary size and recode it so that some data fields represent cases, and some represent variables (see ‘Analyse this’). If you’re analyzing several data sets for comparative trends – sales data from different stores, different counties, and different states on a variety of products, for example, or affirmative-action results in a database of colleges from several regions – this facility is just what you need. The Restructure Data Wizard is unique to SPSS, and it could dictate many users’ buying decisions. A helpful feature of SPSS – the Statistics Coach – illustrates what analyses you might actually want to perform. Since many users may have forgotten which test does what, it’s a nice concession to reality. Heavy-duty data crunching You can use the Restructure Data Wizard to take advantage of a new feature in the Advanced Models module: mixed-level models, which allow optimal forecasting for nested models – a tedious chore without wizard help. Other new features include a percentage-change display in online analytical processing (OLAP) ‘cube’ graphics, so you can instantly see time-by-time or region-by-region changes in data, as well as enhancements to logistic regression (new types of probabilities saved in multinomial logistic regression) and categorical regression. These new features put SPSS 11 for the Mac on a level footing with SPSS 11 for Windows, a leader for PCs. But lurking in the background beneath the new tests and enhancements is a strength of SPSS that dates back to the 1970s: syntax files. When you issue the commands to input, transform, and then analyze data, these commands are recorded in a syntax file. When you use the SPSS-automated Production Facility (included on the SPSS CD as part of the Base), you have a way to process files in the background, generating output files and reports with no active user intervention. This is helpful for long, time-consuming data crunches, or for the generation of output files with thousands of graphs, but it’s practically essential if you must create weekly reports on data from the same sources. SPSS will do it automatically, week after week. Be warned that SPSS 11 is geared toward social-science statistics, with little connection to the physical sciences. Although it has several basic types of interactive graphs, nothing is comparable to the statistics-processing disciplines that chemical engineering requires. There’s no serious provision for modern bootstrap or resampling statistics, aside from a new random-sampling command in the SPSS Database Wizard. And if your main concern is quality-control statistics, you may as well resign yourself to firing up Virtual PC and trying to run Minitab 13.
SPSS is adept at handling and restructuring databases from most sources, and it offers excellent advice on selecting appropriate statistical tests. Base-unit upgrades from version 10 to version 11 are cheap, as are module upgrades. For large-scale statistical problems, SPSS 11 is the best choice available on the Mac.