Factorial ANOVA Flashcards
What is the main limitation of t-tests?
Only compares two means. Have to conduct more than one t-test. Increases family wise error/Type I error.
What are the main advantages of ANOVA?
Can compare lots of conditions at once.
Decreases chances of false positives.
Can be used with more than one IV + can see how they interact.
What is the main ANOVA rationale?
Systematic + unsystematic variation. Want to see how much overall variation is systematic vs unsystematic.
We want systematic to be larger than unsystematic.
What is the difference between within-group and between-groups variation?
Within-groups - not due to our manipulation.
Between-groups - our manipulation + random effects.
How do we know if our experimental manipulation has had an effect?
Between-groups > within-groups.
How do we calculate the effect of our manipulation in an independent ANOVA?
between-groups variance / within-groups variance.
What are the 2 main assumptions for an independent ANOVA?
Parametric test!
Normal distribution.
Interval/ratio level data.
Is normality more of a problem for one-tailed or two-tailed tests?
One-tailed.
What is a unique assumption for independent ANOVAs?
Homogeneity of variance.
Levene’s test.
What is a unique assumption for repeated measures ANOVAs?
Sphericity.
Mauchly’s test.
Variances of differences between conditions are similar.
When is homogeneity of a variance less of a problem?
When group sizes are equal.
What are the two main tests that can be used if homogeneity of variance is violated?
Brown-Forsythe.
Welch’s F.
Should Levene’s test and Mauchly’s test be significant or non-significant for the assumptions to be met?
Non-significant.
When do we use post-hoc tests?
When we have no specific hypothesis.
Use Bonferroni.
What are the main conservative post-hoc tests?
Bonferroni.
Scheffe.
What is the main less conservative post-hoc test?
Turkey HSD.
What post-hoc test do we use when equal variances cannot be assumed?
Games-Howell.
What does a more conservative post-hoc test mean?
Less chance of Type I error.
Higher chance of Type II error.
What should you report for post-hoc tests?
P-value + look at means for direction of effect.
What is a factorial ANOVA?
ANOVA with 2+ IVs.
How do you calculate the number of interactions?
2(to the power of)k - k - 1.
Where k is the number of IVs.
What is an additional interpretation we must make when conducting a factorial ANOVA?
We must look at the interaction graph.
What is a 2x3x3 design?
Three IVs; 1st has 2 levels, 2nd and 3rd have 3 levels.
What is the best way to report an ANOVA (give an example)?
(For a 2x3 design): A 2(Gender: male vs female) x 3(Dosage: low, medium and high) independent design.
What effect size is used for a two-way repeated measures ANOVA?
Partial eta-squared.
When do we use planned comparisons?
When we have a hypothesis.
What statistics can we use if Mauchly’s test is violated?
Greenhouse-Geisser or Huynh-Feldt.
What statistics do we report for planned comparisons?
F, df, p, np-squared.
When is there a significant interaction in the interaction graph?
When the lines are crossing/non-parallel.
If the lines don’t cross then the relationship is non-significant.
What is the interaction graph named?
The means plot.
