Quantitative Methods
0% complete

Chapter 5 | Lesson 05.04

Frame the fair comparison

ANCOVA hypotheses ask whether group differences remain after a continuous covariate has been taken into account. The comparison is about adjusted means, not raw averages.

By the end of this lesson, you can

  • distinguish the null and alternative hypotheses for ANCOVA;
  • name the factor, outcome, and covariate in a hypothesis pair;
  • write hypotheses around adjusted group means; and
  • check whether a hypothesis pair matches the research question.

Core logic

Hypotheses test adjusted means

The covariate is part of both hypotheses because the F-test evaluates the group effect after adjustment. Raw group means do not define the ANCOVA hypothesis.

Null hypothesis, H₀

No adjusted group difference

After controlling for the covariate, the population means on the dependent variable are equal across the groups.

Alternative hypothesis, H₁

An adjusted difference remains

After controlling for the covariate, at least one population mean on the dependent variable differs across the groups.

Baseline variationA continuous covariate may influence the outcome.
Statistical adjustmentCompare groups at a common covariate value.
Adjusted decisionDetermine whether group means still differ.

Quick check

Which statement is a valid ANCOVA null hypothesis?

Choose the statement that compares adjusted means and names the grouping factor, outcome, and covariate.

Model example

Academic enrichment programs

A district compares three programs while controlling for students' pre-program reading skill.

One aligned ANCOVA story

Grouping factorProgram: After-School Tutoring, Saturday Academy, Summer Bridge
OutcomePost-test reading comprehension score
CovariatePretest reading comprehension score
Research question

After controlling for pretest reading comprehension scores, is there a statistically significant difference in students' post-test reading scores across the three programs?

H₀

After controlling for pretest reading comprehension, there is no statistically significant difference in post-test reading scores among After-School Tutoring, Saturday Academy, and Summer Bridge.

H₁

After controlling for pretest reading comprehension, there is a statistically significant difference in post-test reading scores among the three programs.

Adjustment is explicitBoth hypotheses state what is controlled.
The outcome stays continuousThe analysis compares post-test score means.
The factor stays categoricalThe three programs define distinct groups.

Guided practice

Translate the structure to healthcare

A clinic compares three rehabilitation programs on recovery time. Patient age is included as a covariate because older patients may recover more slowly.

IVRehabilitation program: A, B, or C
DVAverage recovery time
CovariatePatient age
H₀After controlling for age, there is no statistically significant difference in average recovery time among Programs A, B, and C.
H₁After controlling for age, there is a statistically significant difference in average recovery time among Programs A, B, and C.
1Control phrase

Begin with “after controlling for” or equivalent wording.

2Continuous outcome

Name the score, time, or measurement whose mean is compared.

3Categorical groups

Identify the factor levels that define the comparison.

4Equality or difference

H₀ states no adjusted difference; H₁ states an adjusted difference.

Find what is missing

Repair an incomplete hypothesis

“After controlling for baseline motivation, there is no difference in student achievement.” What is missing?

Your turn

Write one aligned hypothesis pair

Choose a study from your field with one categorical grouping factor, one continuous outcome, and one continuous covariate measured before or outside the treatment.

Guided writing

Frame the fair comparison

Use the same factor, outcome, and covariate in both statements.

H₀ template: After controlling for [covariate], there is no statistically significant difference in [continuous outcome] among groups defined by [categorical factor].

H₁ template: After controlling for [covariate], there is a statistically significant difference in [continuous outcome] among groups defined by [categorical factor].

FactorOne categorical grouping variable is named.
OutcomeOne continuous dependent variable is named.
CovariateOne continuous control variable is named in both statements.

Bring it together

H₀ assumes no group difference remains after preexisting variation is adjusted. H₁ allows a group difference only when it persists after that adjustment.

Source boundary: printed workbook pages 581-587. The next lesson begins with “Assumptions and How to Check Them - ANCOVA.”