Robust ANOVA on Trimmed Means
Robust ANOVA on trimmed means is a heteroscedastic one-way ANOVA that compares trimmed group means using winsorized variances, remaining valid under outliers, skewness and unequal variances.
절사평균 강건 분산분석은 윈저화 분산과 절사평균으로 집단을 비교하는 이분산 일원분산분석으로, 이상치·치우침·이분산에서도 타당하다.
How do I compare three or more groups when the data have outliers and unequal spreads?
이상치가 있고 분산도 다른 세 집단 이상을 어떻게 비교하나요?
This robust ANOVA, often called Wilcox's t1way, applies Welch's weighting scheme to 20 percent trimmed means with standard errors from winsorized variances. With two groups it reduces to Yuen's test, and with no trimming it equals Welch's ANOVA.
When to use Robust ANOVA on Trimmed Means
- Three or more independent groups are compared with outliers or heavy tails.
- Group variances differ.
- A location comparison robust to non-normality is required.
When to use something else
- Data are close to normal; Welch's ANOVA has more power.
- The outcome is ordinal; use the Kruskal-Wallis test.
Assumptions and requirements
- Observations are independent.
- Each group keeps at least two observations after trimming.
- Trimmed population means are the parameters of interest.
What the analysis reports
- F statistic on trimmed means
- Numerator and adjusted denominator degrees of freedom
- p value
- Group trimmed means
How to report it
A robust one-way ANOVA on 20% trimmed means indicated significant differences among groups, F_t([df1], [df2]) = [F], p = [p].
Related methods
Run Robust ANOVA on Trimmed Means on Quorant
- Data you need
- a table (CSV or Excel) with an outcome variable and a grouping variable
- Price
- 250 credits for one analysis up to 10,000 rows (complexity level 2); larger or wider data costs more. Optional methodologist review: 1200 credits.
- Checking
- Checked by property tests in every deployment
- Before you pay
- A free pre-analysis diagnosis checks your data and gives an exact quote; nothing is deducted until you approve it.