Dunnett's Test

Dunnett's test is a multiple comparison procedure that compares each treatment group mean with a single control group mean while controlling the familywise error rate across those comparisons.

던넷 검정은 여러 처치집단의 평균을 하나의 통제집단 평균과 각각 비교하면서 그 비교 전체의 가족별 오류율을 통제하는 다중비교 절차입니다.

How do I compare several treatment groups against one control group after an ANOVA?

분산분석 후 여러 처치집단을 하나의 통제집단과 어떻게 비교하나요?

Dunnett's test compares every treatment group with a single designated control while keeping the familywise error rate at the chosen alpha level. Because it makes only treatment-versus-control comparisons rather than all pairwise ones, it has more power than Tukey's HSD when the control comparisons are the only ones of interest.

When to use Dunnett's Test

• Several doses or treatments, each compared with a placebo or control group
• A continuous outcome across three or more groups with one designated reference group
• Follow-up to a one-way ANOVA when only control comparisons are planned

When to use something else

• All pairwise group comparisons are of interest: use Tukey's HSD
• Ordinal or clearly non-normal outcomes: use the Kruskal-Wallis test with Dunn's test against the control

Assumptions and requirements

• Control group designated before seeing the data
• Independent observations
• Approximately normal residuals within groups
• Homogeneity of variances across groups

What the analysis reports

• Mean difference of each group from the control
• Adjusted p-values
• Simultaneous confidence intervals for each difference
• Group means and standard deviations

How to report it

Dunnett's test showed that [treatment] differed significantly from the control, mean difference = [value], 95% CI [[value], [value]], p = [value].

Related methods

• Tukey's HSD
• One-Way ANOVA
• Bonferroni Correction

Run Dunnett's Test on Quorant