The main changes were to add the new methods discussed in the 4th edition of the text Biometry by Sokal and Rohlf. Changes were also made to the various hints in the output so that they would refer to Sections and Boxes in the 4th edition rather than just the 3rd.
Changes to the user interface:
Some changes were made to ensure compatibility with Windows 7 but the software should be compatible with other recent versions of Window.
Changes made to the program modules
Basic statistics: added the Shapiro-Wilk statistic to test for normality. Bootstrap estimates for the mean, variance, and coefficient of variation are computed. Both parametric and nonparametric estimates of the confidence intervals are also computed. Jackknife estimates can also be obtained for these statistics.
Ancova: Bootstrap estimates for the intercept, slope, and variance components are computed. Both parametric and nonparametric estimates of the confidence intervals are also computed. Randomization/permutation tests can also be computed.
Comparison of means: no change.
Correlation: improved the plotting of the major axes. Added plotting of the Standard Major Axis. Bootstrap estimates for the correlation coefficient is computed. Both parametric and nonparametric estimates of the confidence intervals are also computed. Randomization test can also be computed.
Factorial anova: no change.
Goodness of fit: no change.
Homogeneity of variances: Added the Levene test and removed the log-anova test. Bootstrap tests for the Fmax, Bartlett, and Leven statistics are computed. Randomization tests can also be computed for these same statistics.
Isotonic regression: no change.
Linear regression: Added bootstrap estimates their confidence intervals of the intercept and slope can be computed. . Randomization tests can also be made for the intercept and slope.
Log-linear analysis: no change.
Logistic regression: no change.
Mann-Whitney U: no change.
Mantel test: no change.
Multiple regression: added bootstrap estimates and confidence intervals for the intercept, partial regression coefficients, and R2 values. Randomization tests can also be applied to these same statistics.
Nested anova: no change.
Nonparametric tests of association:.
Polynomial regression: no change.
RxC test: no change.
Single classification anova: added MLS estimates for the variance component,
Thomas and Hultquist (1978) estimates of the variance component and the intraclass correlation coefficient. Bootstrap estimates and confidence intervals are also computed for these two statistics. In addition, a randomization test can also be made.
Two-way anova: no change.
Non-parametric two-way anova: no change.
Probability calculations:. added Shapiro-Wilk and Dunnett distributions.
New modules
Effect size: a new module added to compute the effect size and its confidence limits for t-tests, model I and model II anovas, Chi square or G-test for goodness of fit for frequency distributions, and for multiple correlation coefficients (which also provide effect sizes for use with correlation and regression).
Estimate n: a complete replacement of prior module to estimate the sample sizes needed for t-tests, anova, correlation, tests of proportions, and for multiple correlation coefficients (which also provide effect sizes for use with regression).
Help file
The format of the help file has been changed to make it easier to use. Descriptions of the new modules and the various additions to existing modules have, of course, been added. A number of the existing descriptions have been updated and made clearer. Some new terms were added to the glossary and new output examples were also added. If printed (a PDF file is included for that purpose), the manual will be over 160 pages.
For more information about the methods included in this software, see the 4th edition of the text Biometry by R. R. Sokal and F. J. Rohlf (2012).