SOFTWARE REVIEW
BARRY RENDER, Feature Editor
Crummer Graduate School of Business, Rollins College

Professor Yurkiewicz's review of Studentware Plus in this issue follows his earlier look at NCSS and Execustat (in the December/January 1993 issue). This SPSS product has its roots in a mainframe package many of us lived with in the 1970s-1980s. Studentware Plus is based on the SPSS version 5.0 DOS product. It requires 640K of RAM and 6.5 MB of hard disk space to run.

SPSS' Studentware Plus

by Jack Yurkiewicz, Pace University

Studentware Plus (we'll call it Studentware for convenience; the Plus refers to the fact that this is the second edition of the software), is a subset of the professional package for DOS. There is a Windows version of SPSS, and the company promises to release a student edition for that product. I reviewed the business version, which offers capabilities traditionally needed for a business statistics course. There is a social science adaptation as well.

Capabilities

The student version can handle much larger data sets than the competition: up to fifty variables and observations limited to disk space. I had no trouble working with a file that had over 30,000 elements. Studentware can find descriptive statistics, including crosstabs, perform hypothesis testing for two sample means, multiple regression analysis, analysis of variance with a large complement of post-hoc tests in the one-way model, some nonparametric tests, and time-series analysis.

The graphic capabilities of SPSS has traditionally been weak, and that is the case also for Studentware. All plots are character, rather than pixel. They look poor, but they do have the advantage of being ASCII files, so any word processor can import them. Line plots for time series data are not connected, and cannot even be plotted on one page in the portrait mode if the data set is relatively moderate in size. Studentware has no statistical quality control capabilities.

Studentware has an excellent time series module, offering a full complement of exponential smoothing models. The program can do all combinations of additive, multiplicative, trend, and seasonality models (e.g., no-trend and no seasonality, linear trend and no seasonality, no-trend with multiplicative seasonality, linear trend with additive seasonality, etc.), including newer damped trend models. Studentware can find the optimal smoothing constants, and gave the most accurate results to my sample data of all the software I tried. Curve fitting is also available, with eleven trend-regression models possible. Box-Jenkins ARIMA models are not available, but you can estimate a regression model with first-order autoregressive errors via the Prais-Winsten, Cochran-Orcutt, and maximum likelihood techniques.

Data Entry

If you want to enter the data at the keyboard, you can use the DATA LIST FREE command, and then type the numbers, separated by blanks. This creates an ASCII file. There is a fixed format also, in which you must specify the number of integer and decimal digits for the data. If you want to see the data when you are done, you need the LIST command again. The whole process is not as desirable as entering data via a spreadsheet, but Studentware reads Lotus and ASCII files effortlessly. If your students have access to a spreadsheet that exports Lotus WK1 files, then I recommend that they do their data entry there and then import it into Studentware with the TRANSLATE command.

Studentware has impressive data transformation capabilities, far more comprehensive than the competition. You use the COMPUTE, CREATE, and IF commands. CREATE produces new variables as a function of existing series, and can give you cumulative sums, differences, lags, leads, fast Fourier transforms, moving averages, and many more.

Ease of Learning and of Use

Originally designed to have the user write the command syntax on a sequence of punch cards that would be executed by the computer, SPSS has not outgrown that mindset. The documentation still talks about "jobs" and a "scratch pad." This scratch pad is a "review editor" or text editor on which you can type the sequence of commands. But Studentware also has a pseudo-menu system from which you can choose the commands from a list and "paste" them into the scratch pad. Figure 1 shows this menu system. The screen is divided into three parts, showing the menu of possible commands on the left, the help section on the right (which gives good explanations and offers examples of the command sytax), and the scratch pad at the bottom. All but the most experienced SPSS user will prefer this method, yet incredibly the 604 page manual barely refers to it (just seven pages in Appendix E!). While not as easy as a true pull-down menu system, it does alleviate the task of knowing the SPSS rigid command syntax. Still, it is easy to make errors, and the program "bombs" if you make a mistake. The resulting error messages do not always make it clear where you went wrong. Appendix A of the manual, a full twenty pages, lists those error messages, and the list is not complete. As such, Studentware is the most difficult of the student programs to learn. However, once you master the system, it fares better when it comes to ease of use.

Documentation

SPSS has long been famous for superb documentation, and the book that comes with Studentware lives up to that reputation. It devotes as much space teaching statistics as it does explaining how to use the software. Filled with many interesting examples, numerous screen shots, and many homework exercises, this book, in the hands of a good instructor who can add some cases, can even be used in lieu of a traditional text for the course. If anything, the manual reads more like a textbook on statistics than a primer on how to use the software.

Summary and Conclusions

In summary, this review looked at the student version of SPSS' Studentware Plus while Part 1 (several months ago) looked at NCSS and Execustat. These three programs are significantly superior to the programs that are usually bundled with textbooks. They have distinct personalities, strong in some areas, weaker in others, but all are at the top of the class. Statistix 4.0 (from Analytical Software, St. Paul, MN) is one other program that falls into this rarified category. All of these programs share one drawback: price. Not that they are expensive, but if you add their cost to that of a typical statistics textbook, the total could easily exceed one hundred dollars. That is the only reason instructors might have in refusing to adopt one of them. On the other hand, if a university has a computer laboratory and supplied the software, then any of these programs would make an excellent choice.

JACK YURKIEWICZ is a professor of management science and is the assistant chairperson of the Management Science Department at the Lubin Graduate School of Business at Pace University, New York. He received his Ph.D. in operations research from Yale University. His current interests include computers and software and their use in the classroom, educational software for children, forecasting, and statistical quality control.