SPSS is simple to install, and with the anticipated CD-ROM version, it will be even easier. This reviewer tested it on both a Pentium with 32 megabytes of RAM and on a 486-50 with 16 megabytes of RAM.

The interface has a spreadsheet look similar to Excel or Lotus, and the most interesting and useful component of the interface is that coded information can be viewed either as the actual code or variable labels (in words). This improves the user's ability to search for missing or miscoded data and/or to understand how the data is categorized.

A fantastic addition to the program is its ability to import data from other programs using open database connectivity (ODBC). SPSS imports data from a variety of other databases with relative ease. Gone are the days of defining the variables by having to know the format of the variable (i.e., alphanumeric, numeric, date, etc.), the number of columns which make up the variable, the location in the data set, etc. Yet, SPSS still permits importing of ASCII data and has a simpler interface for defining the variables.

The authors of the latest version have simplified the task of changing variable definitions by combining all of the definition functions (i.e. type, labels, and missing values) into one place. SPSS has also utilized the right mouse button to provide a menu which allows variables to be defined or inserted.

Analyzing data has never been easier. The user can now simply select a statistical test by its general name. Instead of having to search to find which program runs the statistic, the user can access the menu for choices such as regression, correlation, etc. Sub-menus then allow the user to further specify the type of statistic.

For instance, consider a situation in which the user wants to know the relationship between two variables using correlation statistics. S/He chooses 'correlation' under the statistic menu, and then chooses the type of correlation the analysis requires (i.e. bivariate, partial, etc.). A window pops up with all of the variables listed in the left column. Choosing from this list allows the user to enter the desired variables into the analysis. If the user makes a mistake by including a variable that should not have been included, s/he needs only to highlight the variable and to click the button to move the variable back into the original list. The user can further define the type of correlation by clicking on one of three different statistics, and can then choose to run either a one or two-tail test. SPSS runs the analysis, and the results are sent to the output editor.

The new output navigator is another wonderful addition to this program. For one thing, more than one statistical analysis report can be placed in one output file. The same output file can also have analyses from other data sets. Another added feature is that the output navigator is set up in a two-column format much like the file manager of Windows 3.x/NT or Explorer in Windows95. The left column lists the contents of the output file. The right side shows the actual output (tables, graphs, or charts). With this type of configuration, the user can scroll down the output tree to find a particular analysis.

The new capability of having more than one analysis per output file made it necessary for SPSS to include a function that permits the user to print solely the desired analysis. By highlighting the statistic or variable, selecting 'print', and choosing 'print selection', the user obtains only the specified analysis and nothing more. This saves paper, printers, and time waiting for output.

An output navigator also enables the user to change the font, titles, and to add notes to the actual output. Changes are made directly to the output. What better way to individualize charts and tables into exactly what the user wants! Finally, the output navigator permits the user to edit files by changing the names of specific portions of the output, deleting unwanted portions of the output, or reordering the output.

With all of these new changes, one might think that the speed of the program might be at issue. As previously mentioned, this reviewer tested SPSS at two different processor speeds. Of course, the 486 runs SPSS more slowly than the Pentium, but the speed was still tolerable. The Pentium had SPSS zooming! The latest version runs much faster than the earlier Windows' versions. Mainframes, aided by quick processing times, but slowed by time spent writing the syntax file, might still run the new program faster than the Windows' versions, but not by much.

A colleague of this writer, who is statistician currently using SPSS - mainframe, recently reviewed the program, and she was amazed, (if not overwhelmed) by how easily and quickly analyses can be completed. She said that what was accomplished by using the SPSS menus and clicking on variables (which only took a few minutes) would take her much longer just to type in the syntax information required to run an analysis on the mainframe. She was skeptical about whether SPSS would run the correct statistical test, but when she reviewed the SPSS syntax help file and the notes section of the output (which specifies the syntax of the statistical test), she was satisfied that it was the correct test.

SPSS permits data tables and output files to be saved into different formats including SPSS+PC, fixed ASCII, Lotus, Excel, and several others. This reviewer did not determine whether or not SPSS Version 7 files are backwardly compatible with mainframe versions of SPSS.

Additionally, SPSS Version 7 takes advantage of long file names supported by Windows95/NT. This helps decrease problems trying to remember what file has what data.

Every program has its limitations and unfortunately, SPSS is no exception. Variable names are still limited to 8 characters. If the variable name is longer than 8 characters, and the file is imported from an external source, the variable names are lost and are renamed var001, var002, etc. Variable labels do not import, and must be entered through data definitions. Finally, the output file limits the user's ability to view all of the data in a large correlation matrix; thus, the matrix needs to printed. This reviewer was unable to determine a way of decreasing the size of rows so that more data could fit on the screen.

The entire program might be expensive for the casual user because of all of the possible add-on programs available. The good news is that most statistical analyses are contained in the base program, and the purchase of additional programs may not be necessary. Many colleges, universities, hospitals, etc. have special arrangements with SPSS which might alter the cost for individuals associated with those institutions.