Research and Statistical Support From Academic Computing Services

By Dr. Panu Sittiwong, Research and Statistical Support Manager (panu@ unt.edu)

Academic Computing Services has a team of research consultants to provide research and statistical support to faculty and students at UNT. Our consultants are capable of assisting you in the areas of reseach design, statistical analysis and data analysis. Our mission is to assist researchers on campus in carrying out their projects and/or assisting faculty members in teaching their classes. Our services include providing consultation on the research design, data acquisition, and analysis of the data. The extent to which services will be provided varies by individual projects. At one extreme, the consultant will help with writing the appropriate programs to carry out the analysis required. In another case, we may provide consultation and leave programming tasks to the researcher. In the area of instructional support, we can schedule a special seminar and workshops for your class in the area of statistical computing. While research consulting support will vary from one case to another, statistical analysis and data support are somewhat uniform. This article will outline these two services.

Academic Computing Services has statistical packages available for computer hardware running under various operating systems. This article will point out all of the statistical packages which are available at UNT. It will also make a recommendation as to the software and platform that will fit your needs when you're uncertain about your choice.

Host System Statistical Packages

Academic host system computers consist of an IBM 3081 mainframe compatible computer and a Solbourne minicomputer. Currently, there are two major operating systems available to academic users on the IBM mainframe: VM/CMS/XA and OS/MVS. The Solbourne computer is running on the UNIX operating system. The following table shows statistical packages available on the three platforms.

Microcomputer Statistical Packages

There are a large number of microcomputer packages available at UNT. In fact, every department has adopted particular packages which are appropriate for their needs and disciplines. For example, the College of Business adopted Minitab PC for instructional use. The College of Arts and Sciences, on the other hand, has a site license for RATS PC to be used in their classes.

Academic Computing Services maintains a site license for SAS and SPSS products. Currently, SAS/PC version 6.04, SAS/Windows version 6.08, and SAS/OS2 version 6.08 are available. These products include all of the SAS statistical modules. We also distribute SPSS/PC+ version 5.0; SPSS for Windows version 6.0; and SPSS for Macintosh version 4.0. All product packages are available free of charge to all full-time faculty and staff at UNT, or they can be installed on any Novell Network on campus. Teaching Fellows, Teaching Assistants, and Research Assistants may request to have the software installed on any PC that belongs to the university. SAS/PC and SPSS/PC are also available to students in all of the General Access Microcomputer Labs on campus. The Windows-based products will be available as soon as those labs are equipped with PCs with Windows software.

If you need to have any of the above packages installed on your microcomputer, you may contact one of the Academic Computing Services statistical consultants at (817) 565-2324, or you can mail your request to either PANU on the CC1 file server (CC1:PANU for Word Perfect Office Mail or PANU@cc1.unt.edu on Pegasus mail) or PANU@untvm1 on VM/CMS mail. The software will take a minimum of 20M to a maximum of 100M of hard disk depending on the software you requested and the statistical procedure you want. You need to check your hard disk space to verify that you have enough space for the software since our installer will not be allowed to remove any file from your hard disk.

While SAS and SPSS under microcomputer platforms are fully supported by the Research and Statistical Support team, this is not the case for other microcomputer based packages. Our staff will try their best to provide support for those packages. Microcomputer-based statistical packages, however, are so numerous that it makes it impossible for our staff to know them all.

A Platform for Your Statistical Computing

Most of the time, the platform on which you run your statistical package is selected based on previous experience. This rationale, in some cases, however, is contrary to the changes and development of computers systems. For example, MUSIC/SP will be deinstalled from the ACAD mainframe. This change will require all previous MUSIC/SP users to convert their mainframe work to run on either VM/CMS, UNIX, or microcomputers. This section will present some of the advantages and disadvantages of all of the statistical package platforms which are currently available or will be available at UNT in the near future.

Mainframe, Mini, and Micro

SAS users at UNT have advantages over some other statistical packages since SAS is available on all three platforms. (See Table 1 on page 16.) SPSS users can select between SPSS on CMS, SPSS on OS/MVS, SPSS/PC+, SPSS for Windows, or SPSS for Macintosh.

• General Guidelines for Choosing a Mainframe or Microcomputer: Microcomputers, most of which are the 286 or above machines, are now widely available on campus and most researchers on campus now have a microcomputer in their office. They provide a suitable platform for your statistical computing comparable to the mainframe host system. This doesn't mean that you will no longer need to use the mainframe for your statistical computing. Both platforms provide different advantages and disadvantages. This section will try to point out some of those advantages and disadvantages in order to help you make a selection on the platform that will run the software.
• Microcomputer Platforms: There are several options available to you when you select to use a microcomputer for your statistical computing. You may choose to run the software directly under the MS-DOS operating system, or you may want to invest in MS-Windows, or you may decide to use a Macintosh. Both SAS and SPSS for microcomputers provide several advantages, these include:
  1. Both SPSS and SAS on the microcomputer platform provides a Menu system which you can use to compose your programs. This helps to minimize syntax errors. It is possible that a novice user could compose a program without opening the manual.
  2. Both SAS and SPSS provide excellent on-line help systems. All commands and syntax can be acquired from the help command.
  3. Both systems have procedures that will allow users to incorporate the output into some other microcomputer program. For example, the Table Procedure in SPSS/PC+ or PROC Tabulate in SAS/PC are capable of producing a customized table for your data. The output can be directly included in a word processing program.
  4. Graphic output from both packages can be directly incorporated into WordPerfect or DrawPerfect.
• Both SAS and SPSS for Windows and for the PC platform currently take advantage of the Extended Memory System (XMS) and they are capable of using up to 16M of memory. This helps to eliminate some of the previous shortfalls of the microcomputer-based statistical packages, and makes both have the same capabilities as their mainframe counterparts but with the Graphical User Interface (GUI). Hence, if you have a PC which is based on the 386 microprocessor or above, the Windows-based software provides an excellent platform for you. If you plan to upgrade your microcomputer to take advantage of the statistical packages under Windows, you will need to keep the following recommendations in mind.
  1. Plan to install a minimum of 6M of memory on your microcomputer. Both SAS and SPSS will run with less memory but with a performance degradation.
  2. Plan to install the math coprocessor (80x87) chip on your PC.
  3. If you plan to have the software installed on your hard disk, you will need at least a 120M Hard disk for SPSS for Windows, and a 200M Hard disk if you need a full version of SAS for Windows.
  4. Plan to have an EGA (or above) Display.
  5. Plan to have a mouse installed on your PC.
  One shortfall of the microcomputer- based statistical software is printing, especially if you plan to generate a large output on a dot matrix printer.
• Mainframe Platform: Obviously, the main advantage of running mainframe statistical software is the ability to utilize a large database in your research. Secondly, if you don't have a 286 or above microcomputer, the mainframe will provide a more powerful platform for your needs.
• As mentioned before, statistical software is available either under OS/MVS or VM/CMS, or both. In some cases, you may not have any choice but to run your statistical analysis on the mainframe under one operating system. Where choices are available, we recommend that you run your analysis under the VM/CMS operating system. There are several advantages to running the statistical software on VM/CMS. These advantages include:
  1. All of the above software can be run interactively on VM/CMS. In this environment, you will have a software interface similar to some of the microcomputers.
  2. Since the CMS operating system is interactive, all jobs submitted will be executed immediately.
  3. With the implementation of TCP/IP on VM/CMS, users have the ability to move data between the PC and the mainframe quickly. Therefore, a broader range of options can be selected.
  I didn't discuss the UNIX platform since there is a limited amount of software available on the Solbourne. It is a good platform if you are using SAS or IMSL subroutine jobs which are numerically computation intensive.

Data Services at UNT

The UNT Computing Center has volumes of machine readable data available to all students and faculty members. These data are acquired from several sources including the Inter-university Consortium for Political and Social Research (ICPSR), Department of Labor, Center for Research in Security Prices (CRSP), Standard and Poor's (COMPUSTAT), etc.

• CRSP Data: Once a year, UNT will receive data from the Center for Research in Security Prices. The data include: Monthly NYSE and AMEX Returns and Master file; Daily NYSE and AMEX Returns file; Daily NASDAQ Returns and Master file; Daily, Monthly, Quarterly, and Annual Markets Indices file; and National Market System Securities file.

• COMPUSTAT II Data: Similar to CRSP data, UNT will receive, once a year, updates of COMPUSTAT II data from Standard and Poor's. The current holdings include: Primary, Supplementary, and Tertiary (Industrial and Research) file; Over the Counter file; Bank file; and Price, Dividends and Earnings file. The data are available both quarterly and annually.
  

  Academic Computing Services produces a handout on how to access and use the CRSP and COMPUSTAT data sets. These documents are available at the Computing Center main office (ISB-119).
  

• ICPSR Data: Data from ICPSR constitute the majority of the data archives at UNT. Currently, there are more than 200 data titles available locally at UNT. As a member of the Consortium, UNT students and faculty members can request any data made available from the ICPSR. Data available from ICPSR covers a wide range of subjects and disciplines including Public Opinion surveys, Election studies for the US and foreign countries, the Congressional Roll Call, the General Social Survey, the Health Interview Survey, the Consumer Expenditure Survey, the Goverment Finance, World Economic Indicators, the Population Survey, Census information for the US and foreign countries, EURO-BAROMETER, etc.
  

  Our staff are currently working with the UNT library to include all of the titles of the ICPSR data in the library card catalog system. After the process is done, you will be able to locate the data using the UNT library on-line card catalog system.
  

  If you would like to use or want to find out more about the ICPSR data archives, you can stop by and talk to one of our Statistical Consultants or call the Computing Center at (817) 565-2324. We will help you identify the data that are suitable for your needs. If the data are available locally, then we will need to set up a computer program to access the data. If we need to order data from ICPSR, it may take up to four weeks before you can use the data.
  

• Data from Other Sources: In addition to those public data sources, you can acquire data by yourself. We can handle data on various mediums. If you plan to acquire a secondary data source by yourself, you may want to consult with Academic Computing Services concerning the method and medium to store data.

Conclusion

In conclusion, Academic Computing Services' main charter is to provide services in support of the instructional and research activities of the University. In order to achieve this goal, we provide a variety of services and software to our user community. This article was written in order to provide you with information so that you can select and use the software in support of your research.

If you have problems or questions about this server, please contact me as soon as possible. You can send mail to the following address: