This page was last modified 1 March 2004. It is maintained by Mick Keough
This small program (wilcox.exe), written by Andrew Constable, of the Australian Antarctic Division, is a post hoc procedure in ANCOVA, used to compare groups when slopes are heterogeneous. It runs in a DOS window, and we have provided the file in zipped form. You just need to extract it to any directory, and run it. You will be prompted for various sums of squares, etc., from analyses of your data.
We've just been advised by Tom Schoenfeld that there is a problem with the program when dealing with small sample sizes (n = 3, or thereabouts). We'll go back and check the source code, and post a correction when it's sorted out.
We have provided links to two pieces of software that we find useful for power analyses, GPower, and the software provided by Russ Lenth, of the Univesity of Iowa. Both are free, but you should acknowledge the use software in papers. Lenth has developed software for several years, beginning with PowerPack, through to his current Java applets that allow you to do power analysis on line. He has also created a small add-in for Microsoft Excel, PiFace, which includes the non-central probability distribution functions need to compute power. This add-in allows you to program pretty much any power analysis into excel.
You may also be interested in the review of power analysis software published by Len Thomas and Charles Krebs, although it's now 5 years old.
In Section 6.1.16 (pp. 141-142), we describe hierarchical partitioning, a procedure for decomposing the variability contributed by each predictor variable in a multiple linear regression model. Chris Walsh and Ralph Mac Nally from Monash University have written a program in the R language for doing hierarchical partitioning. This program, along with the R statistical software package required to run it, can be accessed from http://cran.r-project.org/. The program can be cited as: Walsh, C., and Mac Nally, R. 2003. The hier.part package Version 0.5-1. Hierarchical Partioning. Documentation for R (R project for statistical computing).
Two of the best places for new, interesting multivariate approaches are web sites of Marti Anderson and Pierre Legendre. Marti's site includes distance-based redundancy analysis (Chapter 17) and her robust, non-parametric MANOVA approach (Chapter 18). Pierre Legendre's page largely supports his Numerical Ecology book, but also has lots of other useful software.