I am pleased to announce Release 4.2 of the Real Statistics Resource Pack. The new release is now available for free download (Download Resource Pack) for Excel 2007, 2010 and 2013 environments.
The Real Statistics Examples Workbook has been updated for compatibility with the new release and can be downloaded for free.
The Real Statistics website is in the process of being updated to reflect the new features. These changes will be made over the few days.
Release 4.2 contains the following new features:
Two Factor ANOVA Follow-up Testing
A new Two Factor ANOVA Follow-up Testing data analysis tool has been added. This tool provides Simple Effects, Contrasts (planned tests) and Tukey HSD (unplanned tests) for two-dimensional data tables for analysis of main effects, simple effects and interactions.
Enhancement to the One-way ANOVA data analysis tool
An option has been added which provides access to Dunnett’s post-hoc test. This test uses a table of critical values. These can be accessed using the new DCRIT function.
Follow-up tests to Kruskal-Wallis have also been added, including Contrasts, Nemenyi, Dunn’s and Dunett’s tests.
The Kruskal-Wallis test now applies a ties correction factor. A new array function KW_TEST is now available which carries out this test, including a ties correction.
Effect sizes have been added to the results of the Tukey HSD test for one factor ANOVA.
Enhancements to Nonparametric tests
The Nonparametric option of the T Test and Nonparametric Equivalents data analysis tool has been modified so that the results using the normal approximation, critical values table and exact tests are more clearly labeled.
An option has also been added which enables you to add a continuity correction. This covers the Mann-Whitney and Signed-Ranks tests.
The MANN_TEST and SRANK_TEST array functions have also been modified to include a continuity correction option.
Diagonal of hat matrix
A new function DIAGHAT(R1) has been added which calculates the diagonal of the hat matrix (for multiple linear regression). The LEVERAGE function now calls this function instead of the HAT function, which means that the LEVERAGE function can now calculate leverage for very large data sets (previously limited to about 2,200 data elements).
The DURBIN function has been enhanced so that the normal approximation is used when there is no entry in the table of critical values. The normal approximation is based on the property that for large samples of size n, the Durbin-Watson statistic is approximately normally distributed with mean 2 and variance 4/n.
The value of Cohen’s effect size in Contrasts for one-way ANOVA has been corrected.
The value for Roy’s Largest Root has been corrected in the MANOVA data analysis tool.