Every time you read that decaffeinated coffee is raising your cholesterol level, or that fireplace bricks are leaking cancer-causing gas into your house, or that tutoring doesn’t raise SAT scores, you are looking at the summary of an ANOVA (analysis of variance) study. Although the relationship between cause and effect is frequently another matter, the establishment of at least a connection between factors and effects is the domain of ANOVA and similar statistical procedures.
Abacus Concepts is justified in calling this program SuperANOVA. It offers not only an encyclopedic collection of standard methods in statistical inference, but also one of the slickest interfaces to be found on any Mac product. To paraphrase Einstein, everything has been made as simple as possible, but no simpler. Abacus has clearly spent lots of time designing straightforward ways of operating this array of sometimes cumbersome traditional tests.
The Spirit of Professionalism
This package is designed for professional statisticians and researchers, and it approaches its task with commendable thoroughness. It includes ANOVA, ANCOVA, MANOVA, MANCOVA, and regression (simple, multiple, and polynomial). The program also reports all types of sums of squares, residuals, and the eigenvalues and eigenvectors computed during MANOVA.
SuperANOVA will accommodate models as large as memory permits. But be warned that if you use SuperANOVA with only 1MB of RAM, these models will be quite tiny — the application itself takes more than a megabyte and only works on a 1MB system through Abacus’s ingenuity.
SuperANOVA handles a wide assortment of designs: block and incomplete block, Latin squares, nested designs, split-plots, designs with missing cells, and any conceivable repeatedmeasures design. The standard analysis window also includes buttons for post-hoc tests (12 of them, including the relatively new Games-Howell procedure) and contrasts.
Amenities
Using buttons for special tests suggests one of the principal differences between SuperANOVA and the general linear modeling section of giant statistical programs such as Systat. When you call up a data file to begin constructing a new model, all the data elements required for model construction are arranged with a few mouse clicks... The data elements appear in a list at the lower left of the screen. You simply select them and make a choice from the buttons at the upper left to identify elements of the model as dependent variables, individual factors, or interaction effects. This interface style means that you can construct a model with no more typing than it takes to label your data elements and enter numbers. As it happens, the program also provides a data entry environment that assists you in coding and formatting, and lets you import spreadsheet, database, StatView, and tabbed text files.
But the cumulative effect of dozens of thoughtful interface touches (the dialog boxes for recoding and transformation of data are exemplary) is simply to make it easier to construct models on the fly, to test and modify hypotheses, and generally to investigate data in an intelligent way. For example, Abacus Concepts' innovative Do-It file reconstructs all your work, not just the data set and last screen, when you choose Save from the File menu. In practice this means you can reconstruct the entire trail of refinement toward development of a final model.
For those just learning about model construction, Alxicus has packaged 24 models, covering different formats in one-, two-, and three-factor ANOVA; repeated measures tests; Latin squares; and others. The manual not only carefully explains the use of models, but also details the computational methods the program employs.
To enhance the bare statistics for reports. Abacus offers the tear-off drawing tools it developed for StatView II, and some convenient SuperANOVA-specific touches. All output tables can be handled as objects, sized, font modified, and emphasized with the drawing tools before export to a word processing program.
A Two-Dimensional Picture
There is no doubt that SupenVNOVA is the outstanding Mac vehicle for performing most varieties of analysis of variance and presenting the results of the analysis in reports with enhanced tables as their principal mode of decoration. One sometimes feels uncomfortable, however, with the program’s firmly unmodern approach to data. Now that rotatable three-dimensional data plots with all manner of brushing, slicing, and data coloring have become common ways of looking at data on the Mac, it’s possible to argue about the best way to perform exploratory surveys of a data set in the search for connections. Nonetheless, once a model has been selected, it’s fair to say that most statisticians would regard SuperANOVA output as more convincing final support for a model than a set of 3-D color-plot views.
Buy It
If you work with statistics professionally, buy this program. It doesn’t do everything, but it does ANOVA superbly. It’s easy to think of individual research projects or consulting assignments in which the time savings and report quality of SuperANOVA would justify its purchase. You will probably want to own programs that highlight visual analysis of data as well, and you are likely need programs that include other types of statistical tests, but you won’t regret having this one in your repertoire.
Seiter, Charles. (May 1990). SuperANOVA 1.01. Macworld. (pgs. 353, 355).
Abacus Concepts
Applications
out of 10 (0 vote)
26 / 
2020-09-15 / 
4afb3827bbea9211d1c93232bb8885cfaffb996b / 
/ 
Super_Anova_1.0.1.sit
Emulating this? It could probably run under: 
Mini vMac
