STATISTICS

Kaleidagraph

2 November 2007

Synergy Software

Reviewed by Felix Grant

Kaleidagraph, from Synergy Software, is in the ‘missing office suite module’ layer of the data analysis market – vital in extending both direct benefits and expertise development into new user groups, whether those be individuals, small organisations or extended areas of larger ones. Often overlooked by reviewers in favour of bigger packages, it offers flexibility without bloat and makes a vigorous contribution to science: as I write, for instance, Kaleidagraph 3.5 has just been cited as the enabling software in a virology study [1] from a group of prestigious academic sources. Release 3.5 was our last review, since when there have been one minor and one major upgrades.

The point step to 3.6 strengthened the analytic side of a package emphasising exploratory approaches: ANOVA (one and two way, including repeated measures), Bonferroni, Dunnett, Friedman, Holm, Kruskal-Wallis, Tukey, Wilcoxon and Wilcoxon-Mann-Whitney. The interface was refined in several ways, as were graphics features, visualisation handling, and output. Release 4 continues the same development track, with a welcome concentration on cumulative effect of individually subtle changes across the package. Two new plot types join the array, but more significant is the combination of less visible improvements to the ways in all types are used. Enhanced display formats, an extensive development of axis control, rationalisation of dialogs, plus type specific extensions, expand the exploratory control and scope which are available. There are operational adjustments on the analytic side, too, including the addition of partial derivatives to curve fits and three tests (Student-Newman-Keuls, Scheffe, and Fisher's LSD) have been added.

The data sheet has been considerably overhauled, coming into line with general spreadsheet conventions and those more specifically adopted in statistics products, the sum effect being to greatly expand the rich communication capacity. A new toolbar at the top of each window can be shown or hidden according to preference, default behaviour can be set and stored, and the data is (unless a specific decision is made to prevent it) automatically embedded on closure into dependent plots. Discontinuity splits allow simultaneous examination for different regions.

A particular feature of Kaleidagraph is the floating Formula Entry bar, analogous to the cell entry box in packages like Excel, but floating over the workspace. It has its own menus (where correlation has now been added to the Functions drop down), providing focused access to frequently -used manipulative functions independently from the main interface. The usefulness of this feature has been enhanced by modification of the statistics commands, which simplify syntax for calculations across column ranges, cutting the parameter tail of a function by up to 50 per cent as well as increasing clarity.


General managerial behaviour continues to evolve in the same spirit as the core. Macro controls have been combined where they were previously in two dialogs. Plot updating is data set-based by default, unless a selection-based alternative is selected. Group layout files can be saved with embedded plot files, and many import/export options have been drawn into an enhanced Save As menu (menu structures have been similarly rationalised elsewhere). Mask inversion has been promoted from simple macro to multiple column command, and the clipboard, which holds coordinates of points identified from plots, can be made persistent so that sequential selections are combined. I have one small wish of my own for the future: that selecting a new plot type when an existing plot is in use should not replace the old with the new.

Some changes mean that existing users upgrading will have to allow for a little acclimatisation time, but that will be a small price to pay for the aggregate improvement that they bring. This is a valuable upgrade to a good and accessible exploration tool for researchers who want to spend time on their own work rather than on becoming data analysis specialists.

1. Jason D. Graci, et al., Lethal Mutagenesis of Poliovirus Mediated by a Mutagenic Pyrimidine Analogue. Journal of virology, 2007.

Click here to find out more