In the decade since the publication of the first edition of this guide (Levine, 1991), and despite the development of several more specialized statistical techniques, analysis of variance (ANOVA) continues to be the workhorse for many behavioral science researchers. This guide provides instructions and examples for running analyses of variance, as well as several other related statistical tests of significance, with the popular and powerful SPSS statistical software package (SPSS, 2001). Although other computer manuals exist describing the use of SPSS, none of them offer the program statements
required for the more advanced tests in analysis of variance, placing these needed programs out of reach. This manual remedies this situation by providing the needed program statements, thus offering more complete utilization of the computational power of SPSS. All of the programs in this book can be run using any version of SPSS, including the recently released Version 11. (SPSS is currently available for a variety of computer system platforms, including mainframe, Windows, and Macintosh versions.)
SPSS for Windows has two methods by which analyses can be conducted: either through the pull-down menu method, in which you point with and then click the mouse (which is henceforth referred to as point-and-click or PAC), or by writing programs. These programs are called syntax and include the commands and subcommands that tell SPSS what to do. Mainframe applications only use syntax. The personal computer packages for SPSS use both syntax and PAC (the exception being the student version for Windows, which lacks many advanced analyses and does not use syntax).
To be able to describe the full spectrum of available analyses and address the needs of the widest number of users, we focus more heavily on syntax, while still including examples for PAC. An additional reason for stressing syntax rather than PAC is that mistakes in the former are more easily recognized and corrected, assuring the user of the validity of the analysis being performed. The principle motive, however, is that there are useful analyses that cannot be performed through current PAC menus (e.g., simple effects).
PAC methods, however, are not slighted. Generally, these too are fully described (albeit comparatively briefly, as befits their lesser capabilities), typically at the end of each chapter. (An exception is chap. 2, where including PAC methods at the ends of the various subsections, e.g., data entry, data importation, saving data, and printing data, made more sense.) Those users intending to use only PAC methods may choose to go directly to those sections.
There are a number of separate programs included within SPSS that are available for ANOVA analyses. These include the ONEWAY, UNIANOVA, GLM, and MANOVA programs. Although portions of the text cover each of these programs (where appropriate), we chose the MANOVA program for primary explication throughout the book, because we find it maximizes the joint criteria of flexibility, power, and ease of use. We find, for example, that there are no analyses of variance tests
that cannot be conducted one way or another by MANOVA, whereas the same is not true for the other programs. A seeming disadvantage of the MANOVA procedure, however, is that it is the only one currently unavailable through PAC. Because we feel that PAC methods are useful only for the simplest of analyses, this is not viewed as a shortcoming.