Software Package Details

The following software is licensed for use on one or more UNIX servers. General purpose packages, such as SAS and Stata, are more widely available. Other packages have narrow application or are focused within particular disciplines.

• DBMS/COPY - data management conversion utility
• Gauss - matrix oriented mathematical and statistical package
• GLIM4 - generalized linear interactive modelling program
• GQOPT - FORTRAN numerical optimization package
• Limdep - general econometrics modelling program
• Lisrel/Prelis - structural equation modelling programs
• Matlab - matrix based system for scientific and engineering calculations
• Rate - event history modelling program
• RATS - econometrics/time-series analysis software package
• SAS
- general purpose data managment and analysis language
• S-PLUS - general purpose, extensible, matrix-oriented statistical package
• SPSS - general purpose statistical analysis and report writing package
• Stata - general purpose, extensible statistical analysis package
• TSP - repeated measures time series estimation and simulation package

Learning A Statistical Package

1. Use the locator links above to determine where the package you need is available. Apply for an account if necessary.
2. Review the learning tools available. Generally it is best to work through one or more tutorials.
3. Determine the local experts. They can often save you hours of frustration, but be considerate of their time.
4. Keep a notebook with key insights and coding examples. These factoids will help consolidate your knowledge. Comment your code to ensure its meaning can be salvaged at a later date.
5. Strive for a big picture understanding of how the package works. Once you have a conceptual overview, the details tend to rapidly fall into place.
6. Develop a system for organizing your programs and output. Naming conventions are important in this regard.
7. Use test code to work out your ideas on a small scale with a few observations. Most packages have some facility for outputting intermediate results that allow you to trace your logic.

Resolving Problems with Package Use

Using a package invariably leads to problems. You don't understand the program. The software is buggy. Your logic is flawed. The data are improperly structured. The pitfalls are endless...

The following guidelines generally apply:

1. Consult the software documentation. Typically there will be a basic user's guide and additional detailed reference guides. There is no substitute for a thorough reading of the documentation.
2. Consult FAQ (Frequently Asked Questions) documents that cover the common issues or available search engines that might provide a hit.
3. Consult an experienced user.
4. Carefully review the logs produced by the package. Many error messages contain helpful clues that can be deciphered or that give hints about further things to test.
5. Learn to distinguish between syntax errors (caused by illegal coding of statements) and logical errors (caused by flaws in the way you have conceived the computing or modelling process). The former are usually cleared up by a careful reading of a manual. The latter will persist and often require assistance.
6. Before calling for technical support, save digital or hardcopy logs. Support people must have the full information provided by a listing of your code and the messages produced by the package.
7. Have the following information when contacting technical support:
   • the package release - as obtained from a log
   • the license number of your installation - as obtained from a log
   • the operating system version you are running under