Reduce noise
Explain outcome variability associated with a relevant background measure.
ANCOVA · Lesson 05.01 · Pages 558–568
A Fairer Comparison
ANCOVA combines ANOVA and regression to compare group means after accounting for a continuous background variable. The central question is simple: do the groups still differ after everyone is placed on a more comparable starting line?
By the end, you can
- Explain how ANCOVA differs from ANOVA.
- Identify a factor, outcome, and covariate.
- Recognize when a covariate fits the design.
- Describe adjusted means and the key assumptions.
- Interpret and report an adjusted group comparison.
What ANCOVA does
Compare groups after accounting for where people started
Real groups often differ before a program, treatment, or class begins. ANCOVA uses the relationship between a continuous covariate and a continuous outcome to calculate adjusted group means.
Improve precision
Make the group comparison after accounting for a measured starting difference.
Clarify the pattern
See whether the group difference remains, shrinks, or disappears after adjustment.
after controlling for
holding constant
adjusted means
Design setup
Build the three-part ANCOVA story
Use one primary scenario: compare final exam performance across teaching methods while accounting for students' pretest scores.
Teaching method
Lecture, Flipped, or Inquiry. The grouping variable is categorical.
Final exam score
The dependent variable is continuous.
Pretest score
A continuous starting measure that may be related to the final score.
Covariate check
A useful covariate is continuous, measured before or independently of the treatment, and related to the outcome.
Adjusted means
See what changes after adjustment
Raw means describe the observed groups. Adjusted means estimate the group averages after accounting for the covariate's relationship with the outcome.
Before adjustment
Lecture78
Flipped85
Observed gap: 7 points.
After controlling for pretest
Lecture80
Flipped83
Adjusted gap: 3 points.
The smaller gap suggests that some of the original difference was associated with pretest performance. A remaining adjusted difference still needs its ANCOVA F test, p value, effect size, confidence intervals, and design context before drawing a conclusion.
Model wording: After adjusting for pretest scores, teaching method was associated with final exam scores, F(2, 86) = 4.21, p = .018.
Fit and assumptions
The adjustment must match the data
ANCOVA relies on the usual group-comparison assumptions plus a linear covariate relationship and similar regression slopes across groups.
Continuous pretest measured before instruction
It may explain starting differences without being caused by the teaching method.
Categorical outcome or clearly different slopes
Standard ANCOVA no longer answers the intended adjusted-mean question.
Interpret and report
Tell the adjusted comparison clearly
A complete report shows what was controlled, whether the adjusted groups differed, how large the effect was, and how precise the adjusted means were.
| ANOVA | ANCOVA |
|---|---|
| Compares group means | Compares group means adjusted for covariate(s) |
| Asks whether groups differ | Asks whether groups differ after controlling for something else |
| Reports means, SDs, F, p, and effect size | Adds covariate effects, adjusted means with CIs, and assumption checks |
Describe
Raw group means and SDs, plus covariate descriptives.
Test
F, degrees of freedom, p, and partial η² or ω² for the group and covariate effects.
Translate
Adjusted means with 95% CIs and a plain-language statement about what changed after adjustment.
Bring it home
ANCOVA is the signal to compare groups after controlling for a relevant continuous variable. The goal is not to erase differences, but to make the comparison more informative by accounting for a measured source of background variation.
Source boundary: printed workbook pages 558–568. The next lesson begins with “Before You Write: ANCOVA Pre-Flight.”