| Academic Computing and Communications Center | ||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
Statistical Tools of the Trade |
||||
This article was written by the ACCC's consultant for statistical software, Violeta Carrión, STATS@uic.edu. |
||||
| Or a Tale of Many Ss | ||||
|
Numbers, numbers, numbers! To make sense of so many numbers, you can use one of the many available statistical programs. The good news is that the ACCC has site licenses for many of the major statistical packages, including SAS, SPSS, STATA, and S-PLUS. (For a list of ACCC-supported packages and links to information about them, see: http://www.accc.uic.edu/stats/swpages.html) Depending on your area of research, you are probably more familiar with one statistical package or another. Social scientists typically prefer SPSS. Biostatisticians typically prefer SAS. But there are are many newer packages now that attempt to give you the best of all worlds -- a friendly interface, a powerful analytical tool, and a reasonable amount of programming power. What works for you will depend on a combination of various factors: your general comfort with using software, your programming experience, the size of your data, and the complexity of your analysis. Also the habits and needs of those with whom you share your data may influence your choice of software. Some people use spreadsheet or database programs for statistical analysis, but that's like using a soup spoon to serve a stew when a serving spoon is available. Statistical packages offer you optimal tools for data management and analysis. Lets explore what the two major packages, SAS an SPSS, have to offer us. |
||||
| SPSS | ||||
|
SPSS, formerly "Statistical Package for the Social Sciences," is quite popular not only because it is a valuable analysis tool, but also because it's user-friendly and visually oriented. Open SPSS and in a few minutes you can be using its pulldown menus to run an analysis. SPSS originally emphasized its programming language, but soon became "pulldown" inclined via its windowing environment (which dates back to its DOS implementation!). But the SPSS programming language and syntax is still where the real power of this program resides. SPSS, like other statistical programs, has evolved into a set of modules. Most of SPSS's modules are available in its main installation media. Some modules that are SPSS's versions of particular algorithms have been packaged as stand-alone products. The installation CD for SPSS contains the following modules: Windows Base, Professional Statistics, Advanced Statistics, Tables, Trends, Categories, Conjoint, Exact Tests, Missing Value, Smart Viewer, and Maps. All of these modules are available under the ACCC's basic SPSS site license. Also available from the ACCC, as stand-alone products, are: Amos, Data Entry, and AnswerTree. AnswerTree is used for automatically finding statistically significant groups in your data via decision tree algorithms. If you have been around for a while, you might recognize this product under its original name, CHAID. With the new name, a set of new algorithms has been added to this decision tree software. (For more information, see: http://www.spss.com/answertree/) "CHAID, for Chi-square Automatic Interaction Detector ..., is an exploratory method used to study the relationship between a dependent variable and a series of predictor variables. CHAID modeling selects a set of predictors and their interactions that optimally predict the dependent measure. The developed model is a classification tree (or data partitioning tree) that shows how major "types" formed from the independent (predictor or splitter) variables differentially predict a criterion or dependent variable." (From "CHAID," http://www.themeasurementgroup.com/Definitions/chaid.htm) On the other hand, Amos is used for structural equation modeling and analysis. Formerly known as LISREL, SPSS's version of Amos is a competitive package that "implements the general approach to data analysis known as structural modeling, analysis of covariance structures, or causal modeling. This approach includes as special cases many well known conventional techniques, including the general linear model and common factor analysis." (From: "What is Amos?", http://www.smallwaters.com/amos/whatis.html) More detailed information on SPSS products is available from the SPSS site at: http://www.spss.com/products/ |
||||
| SAS | ||||
|
SAS, originally called "Statistical Analysis Software," has generally been preferred by programmers, primarily due to the power of its programming language and the portability of its code. When you open SAS, you get the somewhat uneasy feeling that the pulldown menus are not going to help. If you have no programming experience, it will take a bit longer to warm up to this powerful tool. But as you get better acquainted with the program, you will come to appreciate its power -- one line of SAS code can do many things for your data. Lately, SAS has been paying attention to the fact that not all its users are programmers and has improved its pulldown menus. Yet its most critical improvement has been with a new add-on product called Enterprise Guide. This program installs "on top" of your current installation of SAS, but it can also be individually invoked; running SAS does not imply running Enterprise Guide, but running Enterprise Guide does open SAS. Enterprise Guide features include a visual display full of helpful pulldown menus, project management tools, and a windowing environment that creates the needed SAS code for you. Enterprise Guide is included when you purchase Base SAS under the ACCC site license or upon request from qualified UIC SAS user. SAS has changed a lot across versions, with most of the changes catering to the business community. SAS version 8.2 is the apex of this fact. The version 8.2 installation is noticeably hard drive intensive due to inclusion of a large set of products which are mainly geared towards an "enterprise-level" installation. These are products for a SAS server that might be used at a large company doing e-business, with all the bells and whistles you can imagine. They are a hindrance for the academic user who typically wants to use SAS mainly for its statistical analysis tools. To complicate matters, SAS does not package its software by the needs of the user, but by the products the site has licensed. That's of relevance at UIC because the ACCC has upgraded to SAS's "Gold Tier Academic License." The good news is that this license offers a set of new products not previously available under the UIC site license -- for example, the SAS data mining product Enterprise Miner. The bad news is that when you purchase SAS from the ACCC, you get a CD folder full of media, containing a bundle of products, many of which might not be useful to you. To aid you in the selection process of the SAS products you need, we have opted to routinely include with your SAS purchase only those CDs that are typically used for purposes of academic statistical analyses. If you so desire at the time you purchase Base SAS, you can request any of the other products whose installation media we haven't included. In addition, we have a Web page that will guide you through a custom installation of SAS that is suited to your needs: http://www.accc.uic.edu/stats/sas/v8install.html From the extensive list of SAS products, the ACCC has licensed the following: Base SAS, SAS/AF, SAS/ACCESS, SAS/ASSIST, SAS/CALC, SAS/CONNECT, SAS/EIS, SAS/ETS, SAS/FSP, SAS/GIS, SAS/GRAPH, SAS/IML, SAS/INSIGHT, SAS/LAB, SAS/OR, SAS/QC, SAS/SHARE, SAS/SPECTRAVIEW, SAS/STAT, Online TUTOR, SAS/IntrNet, SAS/MDDB, SAS/Secure, SAS/Access to R3, Scalable Performance Data Server, Enterprise Guide, AppDev Studio, SAS/Warehouse Administrator, Enterprise Miner, IT Service Vision, and Enterprise Reporter. More detailed information on SAS products is available from the SAS site at: http://www.sas.com/products/index.html We hope you now feel properly introduced to SAS and SPSS. Feel free to contact us for statistical software support matters, suggestions about our Web site, statistical seminars, or any statistical support issue.
|
||||
| The A3C Connection, 2001-2002, Number 1 | Previous: Figure 2: Quota, Download, Delete Tool for mailserv | Next: Disk Space on borg |
| 2002-6-4 connect@uic.edu |
|