Analysis of Variance (ANOVA)
General resources
A tutorial geared towards psychology research.
One Way
Two Way
Repeated Measures
In R, repeated measures ANOVA needs to be fit using aov, so that a multi-stratum error term can be a part of the model (lm does not support this). lme (from the nlme library'') can also be used for this purpose, specifying the error terms in the random= argument.
- A compendium of resources
- http://statistics.ats.ucla.edu/stat/r/seminars/Repeated_Measures/repeated_measures.htm
Post-hoc analysis
If multiple treatments need to be compared after the experiment, a couple types of post-hoc analyses can be used to test whether the means of the treatments are significantly different. These are the Tukey range test (also known as Tukey-Kramer, or Tukey HSD) and Scheffe's method.