ProStat 4.8 is a recent new release statistical toolbox from Poly Software International. Since we haven’t reviewed products from this company before, it made sense to also look at the closely related PSI-Plot currently in release 8.8.
The packages share a large and useful core of functional equivalence, their differences being in the specifics of application. This means that both could be effectively deployed within a single organisation, giving different users optimum tools, with confidence that experience on one would transfer in large measure to use of the other. They share a common and standard Windows compliant GUI, a large spreadsheet with manipulative functions, batch and template options plus user editable hot keys, a capable (and pleasing to use) object-based graphics editor with output to most file formats, and flexible customisation of layout and output. Saving results back to the worksheet, and 3D interpolated mesh generation are common to both. Such similarities continue throughout.
The differences are generally in extent or depth of a tool class rather than presence or absence. ANOVA in PSI-Plot, for example, becomes ANOVA, ANCOVA and MANOVA in ProStat while the very useful range of plot types in ProStat is extended in PSI-Plot by the addition of options such as Pareto charts, Smith curves, ternary, vector, and column plots. FFT is available in both, but extended differently according to the likely emphases of the different target users, and so on.Both are easy to learn and use, and there is good set of tutorials is provided. I was productive in a couple of minutes, and a guinea pig with no prior statistics software experience produced useful graphics and results after half an hour with no support other than the built in help system.
Focusing on new aspects of ProStat in release 4.8, for existing users there is first of all a solid upgrade of the core statistical capability. Multiple Kruskal Wallis comparisons, nonlinear regression enhanced by addition of simplex and Powell methods, and dose/response calculation and confidence at arbitrary percentage probability levels, a substantial expansion of predefined regression models (use defined models can always be added) to name four examples.
The worksheet gains improved editing operations, new display formats, native handling of wave data, statistical variable descriptors, sorting on multiple keys, evaluation of functions, and hypothesised interpolation of missing data points. Maximum sheet size is now a very useful 1024 variables over 524,288 cases which, while not as large as a Quattro Pro or Excel 2007 spreadsheet, is more than ample for most target users of such a package. It is also, of course, four times the variable count and eight times the case capacity of the default market standard pre-2007 XLS sheet. There are numerous detailed developments in both interface and program, as well, including Windows Vista readiness, although compatibility is preserved back to Windows 95 (I checked this as far as Windows Me).
Some additions increase the interoperability between the products – candle stick plots, for instance, arrive from PSI-Plot and are only one of several new graphic visualisation types. Graphic handling is also enhanced in a variety of ways, both operationally and visually.
For the many science areas, locations and users for which an exploratory, task-focused, work station level tool is more appropriate and more productive than a corporate level package, either of these would be an excellent choice. For those seeking such a solution, both packages should be examined in relation to task differentiation; for those who already have one or the other, the new release developments make consideration of an upgrade well worthwhile.