Two variables that appear related, and a third that might explain the whole thing. What holding a variable constant actually does, what it cannot do, and how to report it in APA 7.
A partial correlation measures the relationship between two variables after removing what a third variable explains in each of them. The classic illustration: ice cream sales correlate with drowning deaths, and both are driven by temperature. Hold temperature constant and the relationship collapses.
Mechanically, it correlates the residuals. It predicts each of your two variables from the control variable, keeps what the control could not explain, and correlates those leftovers. If the original relationship was entirely a shared dependence on the control, the partial correlation drops toward zero.
Comparing the raw and partial coefficients is where the interpretation lives. A relationship that barely moves survives the control. One that collapses did not exist independently of it.
It only controls for what you measured. This is the limit that matters most. Holding one variable constant says nothing about the ones you did not collect, and there is always another candidate. A partial correlation reduces the list of alternative explanations; it does not close it.
The control variable must be measured well. An unreliable control under-adjusts, leaving part of the confound in the residuals and making the partial correlation look larger than it should. The apparent survival of a relationship can be a measurement artifact.
Do not control for something on the causal path. If the third variable is a mechanism through which the first affects the second, removing it removes the very effect you are studying, and the partial correlation understates a real relationship. Deciding this requires theory, not statistics.
It is still not causation. A partial correlation is an association with one alternative explanation subtracted.
Bring in the data. Import from CSV, Excel, SPSS .sav, Stata .dta, or a ReliCheck survey package.
Configure the two variables of interest and the control variable in the Setup panel on the right, and the result appears in the center panel.
Run the unadjusted correlation first. The partial coefficient is only interpretable next to the raw one, because the story is in how much it moved.
Watch the degrees of freedom. Each control variable costs one, which matters in a small sample and changes the p value.
Report both coefficients, name the control variable, and give the partial's degrees of freedom.
Workload and burnout were correlated, r(248) = .46, p < .001. Controlling for years of experience, the relationship remained, rp(247) = .41, p < .001.
Degrees of freedom are n − 2 − the number of control variables. One control on 250 cases gives 247.
Say why you chose that control variable, and name the confounders you could not measure. A partial correlation that omits its own limits reads as a stronger claim than it is.
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 correlation and nonparametric procedures are validated against R's cor and against hand-computed closed-form values, matched exactly. The engine is pure Swift with no R or Python at runtime. Full record on the validation page.
Correlation 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.
Pearson the unadjusted relationship · Spearman for ordinal data · ANCOVA group differences with a covariate held constant · the complete list on the analyses page.