A Quick Look at SPSS Version 11   By Frank LoPresti frank.lopresti@nyu.edu

What's the SPSS 11 story?

The NYU Computer Store sells SPSS 11 at a reasonable student price but most Version 10 owners probably don't need to spend another couple of hundred dollars to move from Version 10 to 11. Unlike the researcher's site license, which is an annual license that costs $100 the first year and $30 each year thereafter, the student's one-time purchase does not include upgrades.

Fig. 1 One schema for rater data--one line for each rater's score for each skater.

Versions 7, 8 and 9 of SPSS were fairly similar--they all had the same modern feel and look. The major enhancement in SPSS 10 was the new Data Editor (see my article, "New Data Editor in Version 10 of SPSS," available online at www.nyu.edu/its/socsci/Docs/SPSSv10ed.pdf). SPSS 10 was a substantial improvement over Version 9. The innovations in Version 11 compared to Version 10 are not as significant as the previous upgrade. The Version 10 user won't notice much of a difference. Unless you're interested in Linear Mixed Models, there are no changes that would make most Version 10 handouts outdated.

SPSS 11 will run on any OS (Windows 98, 2000, NT, etc.) and machine that is currently running Version 10. XP, however, is happier with Version 11.

New to SPSS Version 11

Fig. 2 Second schema--one line of data for each skater.

Let's use the example of rating figure skaters using four raters from the U.S., Canada, Russia and France. A "Two-Way Analysis of Variance" requires a spreadsheet with one line of data for each rater's judgment of each skater (see fig. 1).

The "Inter-rater Reliability" procedure, on the other hand, requires the data to be in a different format with one line of data per skater (see fig. 2; see also Bob Yaffee's article, "Enhancements of Reliability Analysis: Application of Intraclass Correlations with SPSS/Windows v.8," available online at: www.nyu.edu/its/socsci/Docs/intracls.html).

Moving between these two schemas is quickly done using the new "Restructure" command. Today we often store data in relational databases we've designed as "normalized", with no repeated fields. This normalized design has many advantages, one being the knowledge that the database may be redesigned without having to re-input data. The Restructure Wizard gives us the functionality to structure our data for any analysis by configuring our data in many array schemes. Along with my old favorite, "Aggregate", "Restructure" will be a powerful data programming tool in my toolbox.

Fig. 3 When saving you may now select which variables to keep.

The SAVE parameter "/KEEP" has been available to syntax users in previous versions of SPSS. When saving, the "/KEEP" parameter permits saving some, not necessarily all, of the variables in a data set to create a smaller, more manageable data set.

Say you start your analysis with a dataset made up of 60 items forming 5 different scales. First, you analyze the 60 items--missing values, reliability. Next, you compute the 5 scales. Now you want to create a dataset to save without the items--only the demographics and the scales. The 60 separate items are no longer important.

Fig. 4 Pull down from "Analyze" then "Descriptives" for Ratio Statistics.

In Version 11, in the pull-down menu "File", the "Save" file now gives the user the option to select variables to be saved (see fig. 3). The Database Wizard opens up the world of relational databases to SPSS programmers. The SPSS Database Wizard makes writing SQL for retrievals possible for non-SQL programmers, though a bit of reading about relational databases is useful. Pull-down menus allow the SPSS programmer to retrieve from the Microsoft Data Access pack (especially Microsoft Access); Sybase 11 and 12; Infomix 7.3+, 9.14; Infomix 2000 (9.20); UDB (DB2 6.1 and 7.1); SQL Server 2000; Oracle 8.06; and Oracle-8i Releases 2 and 3 (8.1.6, 8.1.7). See the Spring 2000 Connect article in which I discuss using SPSS as a conduit into and out of Microsoft Access ( www.nyu.edu/its/pubs/connect/archives/00spring/loprestidatabases.html).

Improved reading from current versions of SAS data files and SAS portable files. The "Open" command under "File" now provides for a larger choice of SAS file types.
 

 

Fig. 5 Ratio analysis.

The "Ratio Statistics" procedure provides descriptive statistics for ratios between two variables. This new procedure allows you to designate two scales and a grouping variable. It calculates the ratio and describes the ratio over the grouping variable. For example, using the car data provided by SPSS, we can calculate the ratio of horsepower over car weight and designate the grouping variable country of origin. We get an analysis of the ratio by country of origin (see fig. 4 and fig. 5).
 

Summary