- The omnibus F test used in one-way ANOVA asks whether there are any differences between groups. This protects us from the Type I error we might encounter if we tested all possible paired differences in the study design.
- Formulas for ANOVA are based on a structural model that maps sample scores in terms of the population, group, and individual scores (X = μ + α + ε).
- To calculate our F test, we have to partition the scores in the data set into an estimate of between-groups variance (how far groups are from everybody) and an estimate of within-groups variance (how far individual scores are from the group). The F test is the ratio of the between-groups variance to within-groups variance.
- Once you have determined that there is a difference (reject the null), you do either planned or unplanned comparisons to locate where specific differences are between the study groups.
- Effect sizes (η2 or ω2) estimate the size of the difference between groups.
Test your knowledge of this workshop by taking the Workshop Quiz.
|
|
|
