Group differences with a covariate held constant. What adjustment actually does, the two assumptions that decide whether the result is interpretable, and how to report it in APA 7.
ANCOVA compares the mean of one numeric outcome across groups while statistically adjusting for one or more continuous covariates. It answers the question: if every group had started from the same place on the covariate, would their outcomes still differ?
The most common use is a pre-test. Three classrooms are compared on a final exam, but they did not begin equal, and the pre-test explains part of the final score. ANCOVA removes the portion of the outcome the covariate accounts for, then compares what is left.
It does two useful things at once. It adjusts for a baseline difference the design could not prevent, and it soaks up variance that would otherwise sit in the error term, making a real group effect easier to detect. That second benefit is why ANCOVA is worth running even in a randomized study where groups start equal by design.
The result arrives on four tabs: Adjusted means, Group effect, Covariates, and Reporting.
Adjusted means is the tab to read first, and it is the one most software buries. It puts raw M and adjusted M side by side for every group, with n. The gap between the two columns is what the covariate did. A group whose adjusted mean barely moves was already typical on the covariate; one that shifts noticeably was not.
Group effect gives the adjusted F with both degrees of freedom, p, partial eta squared with a magnitude word, and n as complete cases. Read the case count: ANCOVA needs the outcome, the grouping variable, and every covariate present, so a case missing one covariate is dropped entirely.
Covariates gives each adjustment term with b, t, and p. This tells you whether the adjustment did anything. A covariate with a large t is carrying real variance; one near zero is costing you a degree of freedom for nothing.
Reporting writes the finding in APA style and, importantly, carries an assumption prompt with it.
Homogeneity of regression slopes. The relationship between covariate and outcome must run in roughly the same direction and steepness within every group. If the pre-test predicts the final exam strongly in one classroom and weakly in another, there is no single adjustment that is fair to all of them, and an adjusted group difference is not interpretable.
Quanta states this on the Reporting tab rather than leaving it to your memory: ANCOVA assumes the covariate-outcome slope is similar across groups, and it tells you to inspect that before interpreting the adjusted means. Most packages hand you adjusted means with no such warning, which is how an uninterpretable adjustment reaches a manuscript.
The covariate must not be affected by the treatment. Measure it before the intervention, or choose something the intervention cannot change. Adjusting for a variable the treatment itself moved removes part of the effect you are trying to measure. A pre-test taken before assignment is safe. A mid-programme measure is not.
Watch small groups. An adjusted mean computed from three cases is not a stable estimate, however precisely it is printed. Read the n column beside the adjusted means before you interpret any group, and consider collapsing or excluding a group that thin.
The covariate should be measured well. An unreliable covariate under-adjusts, leaving some baseline difference in the outcome. This is one reason a reliability estimate for a covariate scale belongs in the same paper as the ANCOVA. See Cronbach's alpha.
Lay the data out one row per person: one column for the group, one for the numeric outcome, and one per covariate.
Open Compare Means and choose ANCOVA. Compare Means holds four analyses on one segmented control: t-Test, Paired t-Test, One-way ANOVA, and ANCOVA. Choosing ANCOVA changes what the Setup panel asks for, since it now needs covariates as well as an outcome and a grouping variable.
Configure the variables on the right and the four result tabs appear in the center panel.
Check your covariates are correctly scored. A reverse-worded scale item entered as a covariate without rescoring adjusts in the wrong direction. See reverse-coded items.
Use the action bar to add the result to a report, copy the table, or copy the APA text. Every result carries the ReliCheck Intelligence card for a plain-language explanation computed on your Mac.
Report the covariate-adjusted F for the group effect with both degrees of freedom, p, and partial eta squared.
An ANCOVA compared job performance across age group adjusting for four self-awareness items. The adjusted group effect was not statistically significant, F(5, 240) = 0.73, p = .605, partial η² = .015.
Report adjusted means with the raw means available for comparison, not adjusted means alone and not raw means alone. Quanta prints both columns because the difference between them is part of the finding.
State that you checked homogeneity of regression slopes, and say so even when it passed. It is the assumption most often skipped and the one a methods reviewer is most likely to ask about.
Report which covariates were included and whether they were related to the outcome. A reader cannot judge an adjusted effect without knowing what it was adjusted for.
An adjusted difference is still an association. Statistical adjustment is not experimental control, and observational ANCOVA does not license a causal claim.
Every result carries a ReliCheck Intelligence card with one action: explain this result in plain language. It runs entirely on your Mac with Apple Intelligence. No upload, no API key, nothing to configure. The model only explains the result Quanta has already computed and validated. It never computes a statistic, and it never applies a decision on your behalf.
From the action bar you can add the result to a report, copy the table, or copy the APA text. Exported reports carry the APA table, its note, and a caution specific to that analysis beneath it.
Quanta's t tests, ANOVA, and ANCOVA are validated against R's t.test, aov, and car, matched exactly. The engine is pure Swift with no R or Python at runtime. Full record on the validation page.
Compare Means is part of every Quanta subscription, $9.99 a month or $79.99 a year after the 30-day trial. One price for everyone. See pricing.
The analysis, the report builder, and the on-device explanations all work with the network off. Your dataset is never uploaded.
One-way ANOVA without a covariate · Multiple linear regression the model ANCOVA is a special case of · Independent-samples t test for two groups, no covariate · the complete list on the analyses page.