Wilkes Lambda vs Greenhouse-Geisser

I'm trying to use a repeated measures ANOVA, but I'm confused about the difference between the Greenhouse-Geisser and Wilkes Lambda stats. Sometimes the program I'm using gives me both and sometimes only one, and all my data sets are similar. Which one should I use and what is the difference between them?
Hi fowlermel,

Welcome to TalkStats. Below is a tutorial from stats-consult.com. (http://www.stats-consult.com/tutorial-07/tutorial-07.htm). Hopefully it answers your question.

Use the following procedure for interpreting the output:

1. If the multivariate result is used, check the significance value for the Wilkes Lamba statistic: if significant, then go to step 6, otherwise we're done and there are no significant treatment effects.

2. If univariate or alternate univariate result is used, check the Mauchly's Test of Sphericity result.

3. If sphericity can be assumed, we use the univariate statistic (first row of the Tests of Within Subjects Effects): if significant, then go to step 6.

4. If sphericity cannot be assumed, and the univariate statistic (assuming sphericity) is significant, then we use the Greenhouse-Geisser result: if significant go to step 6.

5. If the Greenhouse-Geisser result is non-significant, use the Huynh-Feldt result: if significant, got to step 6, otherwise we're done and there are no significant treatment effects.

6. If the repeated factors are rank ordered (such as time or other incremental variable), the Polynomial Contrasts table can be interpreted: check for significant linear, quadratic or cubic trends in the levels of the repeated factor.

7. If desired, multiple Paired Samples T-Tests can be conducted to compare differences between repeated levels.

8. Remember to guard for type I errors if running multiple paired samples t-tests: if there are 6 or more desired comparisons, we can chose to use a Holms sequential Bonferonni correction, otherwise use a straight Bonferonni correction.