ANCOVA

ANCOVA (analysis of covariance) is a linear model that compares group means on a continuous outcome while statistically adjusting for one or more continuous covariates, such as a baseline score.

공분산분석(ANCOVA)은 사전 점수와 같은 하나 이상의 연속형 공변량을 통계적으로 통제한 상태에서 연속형 결과변수의 집단 평균을 비교하는 선형모형이다.

How do I compare group means while controlling for a baseline score or another covariate?

사전 점수나 다른 공변량을 통제하면서 집단 평균을 어떻게 비교하나요?

ANCOVA compares adjusted group means on a continuous outcome after removing the linear effect of one or more covariates. It fits designs such as pretest–posttest comparisons, where adjusting for the baseline increases precision and accounts for pre-existing differences.

When to use ANCOVA

• Comparing treatment groups on a posttest score while adjusting for the pretest
• A continuous outcome, a categorical grouping factor and one or more continuous covariates
• Increasing statistical power by removing covariate-related error variance

When to use something else

• Covariate effects differ across groups (non-parallel slopes): use moderated regression with interaction terms
• Outcomes are repeated over several time points: use a linear mixed model

Assumptions and requirements

• Linear relationship between covariate and outcome
• Homogeneity of regression slopes across groups
• Normally distributed residuals with equal variances
• Covariate measured without being affected by the treatment

What the analysis reports

• F test for the group effect adjusted for covariates
• Partial eta squared
• Estimated marginal (adjusted) means
• Covariate coefficients
• Pairwise comparisons of adjusted means

How to report it

After controlling for [covariate], there was a significant effect of group on [outcome], F([df1], [df2]) = [value], p = [value], partial η² = [value].

Related methods

• One way ANOVA
• Multiple Linear Regression
• Linear Mixed Model

Run ANCOVA on Quorant