statistiXL 1.5 Released
- A new module has been added to Regression allowing for the comparison of multiple simple regression equations. The module initially compares the slopes of the individual regressions, with the option to perform post-hoc multiple comparison tests if differences between the slopes are found. If no differences in slope are found then a overall test for differences in elevation of the regression lines is performed. Again, there is an option to perform multiple comparison tests if differences are found. If however, the slopes of the individual equations are not similar, i.e. significant differences between them are found, an overall test for elevation is not possible. In this case, it is possible to test for differences in elevation at a specified value of X.
- It is now possible to perform a Cluster Analysis by inputting a pre-determined Similarity Matrix. In previous version, only a Distance Matrix was supported.
- Correlation analyses now outputs the test statistic used in determining the probability of a given correlation (t for simple and partial correlations, F for multiple correlations).
- An additional Default Style has been added to the Options/Screen Preferences dialog. It is now possible to choose between the default Dark / Light Green style and a new Blue / Yellow style.
- Discriminant Analysis can now analyse datasets with more than 32000 cases.
- Cluster analysis now supports more than 255 items. Previous version of statistiXL were limited to 255 items due to the 256 column limit within Excel. statistiXL 1.5 overcomes this limitation by splitting large table into multiple sections.
- Scatterplots in Discriminant and Principal Component Analysis can now include more than 12 Coefficients on the plot.
- The Kolmogorov-Smirnov statistic in Goodness-of-Fit tests now supports larger sample sizes.
- Prior to this version, statistiXL sometimes incorrectly reported the overall (total) proportion of cases correctly classified in Discriminant Analysis - Classification. The value reported was only correct when the number of cases in each group was constant. statistiXL now reports the correct proportion for analyses with uneven groups.
- The regression modules now trap and report the presence of 'constant' variables with-in the analysis (eg a Y Variable having the same value for all measured Xs). Prior to Version 1.5 this simply caused the analysis to fail with a generic "Please Check Your Data" error.
- The Clustering module now traps and reports the presence of 'constant' variables with-in the analysis (eg a Variable having the same value for all cases). Prior to Version 1.5 this simply caused the analysis to fail with a generic "Please Check Your Data" error.
- Prior to Version 1.5 Circular Descriptive Statistics presented an incorrect Mean Angle when the true mean fell between 180 and 270Â°. This has now been corrected.