Tests Flashcards
For what post hoc tests are used for?
Main effects significant.
Used to determine which pairs of means are significantly different from one another.
Cite 2 post hoc tests
Sheffé & Tukey’s Honestly Significant Difference (HSD) test
Which one is the most conservative between Sheffé & Tukey’s Honestly Significant Difference (HSD) test?
Sheffé
Pros of Sheffé’s test
Pros:
- Can be used if groups have different sample sizes
- Less sensitive to departures from the assumption of normality and equal variances in the population
- Most conservative test
Cons of Sheffé’s test
Has lower power to detect differences
How does the Sheffé’s test work?
Compared Fcomparison (that you calculated) with F crit x (k-1)
When is the Tukey’s HSD test used?
If groups have equal sample sizes.
What test used the studentized range statistic, Q?
Tukey’s HSD
How does Tukey’s HSD work?
Calculate observed Q and Q crit (N-k). Compare the two.
What is the H0 in Tukey’s HSD?
H0 : μg = μg′
HSD
Minimum absolute difference between two means required for a statistically significant difference.
What post hoc test do you use when sample sized are unequal?
Tukey-Kramer test
When do you use simple main effects?
When interaction significant
What follow up analysis do you do if:
Interaction significant BUT at least 1 of the main effect is.
Post hoc testing. Comparison of marginal means of each factor.
What follow up analysis do you do if:
Interaction significant BUT dominated by main effect.
Understand main effect first and then evaluate how it changes at the levels of other factor.
What follow up analysis do you do if:
Interaction significant & dominates main effect.
Simple main effects.
What are the heuristics to choose which factor you’re using in a simple main effect? (4)
1) Factor that has the highest number of levels
2) Quantitative factor
3) Factor with the greatest SS for main effect
4) Experimentally manipulated factor
When do you do simple comparisons?
Follow-up analyses for significant Simple main effects.
Comparisons among means of the factor for which we identified a significant main effect.
What is a recommendation for simple main effects and simple comparisons?
Set family wise Type 1 error to 0.10.
Adjust Type 1 error.
When do you need to adjust the alpha family wise? (for main effects)
When only main effects are significant.
Which test correct alpha family wise (main effects)?
Bonferroni’s correction
What is Bonferroni’s correction? (calculation)
apha(FW) = alpha(FW) / number of post hoc comparisons (e.g. 3 if 3 levels)
Controlling Type 1 error when conducing follow-up analyses for significant MAIN EFFECTS.
Bonferroni’s correction
Controlling Type 1 error when conducting follow-up analyses for significant INTERACTIONS
The recommendation in Keppel and Wickens is to set αF W = 0.10 for the set of simple main effects because these effects incorporate variation due to both the main effect of the factor for which we are computing simple main effects and the interaction.
Two types of planned comparisons
Orthogonal (independent) and non-orthogonal (dependent).
When do you do planned comparisons?
Designed before data is collected. Specific hypotheses.