10.5
Mehrfachvergleichstests
View the full transcript and gain access to JoVE Core videos
Q1: What is a multiple comparison test and when should you use it?
A multiple comparison test (MCT) is a post hoc analysis conducted after comparing multiple samples using hypothesis tests like ANOVA. It identifies which specific group differs significantly from others or which factor causes a significant effect. MCT is essential when comparing many groups, as pairwise comparisons increase Type-I error rates, making it difficult to identify truly different groups without statistical correction.
Q2: Why does Type-I error increase when comparing multiple groups?
As the number of sample groups increases, the number of pairwise comparisons grows exponentially. Each comparison carries a 0.05 significance level risk of Type-I error. With many comparisons, these error rates accumulate, increasing the overall probability of incorrectly rejecting a true null hypothesis. Multiple comparison tests correct alpha values to control this cumulative error.
Q3: How do multiple comparison tests reduce Type-I error?
Multiple comparison tests correct the significance alpha values to reduce Type-I error across all pairwise comparisons. By adjusting the threshold for statistical significance downward, MCTs make it harder to incorrectly reject the null hypothesis. This correction ensures that the overall error rate remains at the desired level, typically 0.05, even when conducting many simultaneous comparisons.
Q4: Can multiple comparison tests handle both equal and unequal sample sizes?
Yes, different types of multiple comparison tests are available for datasets with equal or unequal sample sizes. Researchers can select an appropriate MCT based on their data structure. The Bonferroni test is one of the most commonly used MCTs and can accommodate various sample size configurations, making it flexible for diverse experimental designs.
Q5: What is the difference between comparing two groups versus multiple groups?
Comparing two groups requires only one pairwise comparison at a 0.05 significance level, making it straightforward to identify differences. With multiple groups, the number of comparisons increases dramatically, raising Type-I error rates and making it difficult to pinpoint which group is significantly different. Multiple comparison tests address this complexity by correcting alpha values across all comparisons.
Q6: How does ANOVA relate to multiple comparison tests?
ANOVA is a hypothesis test used to compare multiple samples initially. If ANOVA results are significant, multiple comparison tests are then applied as post hoc analysis to identify which specific groups differ. MCTs follow ANOVA to pinpoint the source of significant differences and control Type-I error across pairwise comparisons of individual groups.
Q7: What practical example demonstrates why multiple comparison tests are necessary?
When comparing zebrafish mean lengths across many treatment groups, identifying the significantly different group becomes difficult without correction. With two groups, a 0.05 alpha level easily detects differences. As group numbers increase, pairwise comparisons multiply, inflating error rates. Multiple comparison tests correct these alpha values, reliably identifying which group has significantly different mean length.