A partial correlation measures the relationship between two variables after removing the influence of one or more other variables. It answers questions like: do students who study more score higher on the exam even among students with the same prior grades? The write-up shows the ordinary (zero-order) correlation and the partial one side by side, so the reader sees how much the control variable explained.
What you need before you write a single word
- The zero-order correlation (r, df, p).
- The partial correlation and the variable(s) controlled for.
- Its degrees of freedom: N − 2 − the number of control variables.
- The exact p value and ideally the 95% CI.
The APA 7 format template
[Variable 1] and [variable 2] were positively correlated, r(df) = .XX, p = .XXX. After controlling for [covariate], the partial correlation was [weaker but still significant], r(df) = .XX, p = .XXX, 95% CI [.XX, .XX].
All 53 templates, ready to paste with the italics set: the Rescue Vault, $27 →
Formatting rules reviewers actually check:
- Name the control variable(s) in the sentence.
- Use the right df: one fewer for each control variable (here 100 − 2 − 1 = 97).
- No leading zero on r, p or the CI limits.
A worked example
For 100 students, you correlated weekly study hours with final exam score, controlling for prior GPA.
- Zero-order: r(98) = .56, p < .001, 95% CI [.41, .68]
- Partial, controlling for GPA: r(97) = .27, p = .008, 95% CI [.07, .44]
Written up in APA 7:
Weekly study hours were strongly correlated with exam score, r(98) = .56, p < .001, 95% CI [.41, .68]. After controlling for prior GPA, the partial correlation was weaker but remained significant, r(97) = .27, p = .008, 95% CI [.07, .44], indicating that study time was related to exam performance beyond students' earlier achievement.
Mistakes reviewers catch
- Reporting only the partial correlation. Without the zero-order r, the reader cannot see what the control changed.
- The zero-order df on the partial r. Each control variable costs a degree of freedom.
- Implying causation. Controlling for one variable does not rule out others.
Let KyroStat do the write-up for you
KyroStat runs the zero-order and partial correlations together and writes both into one APA 7 paragraph.
Frequently asked questions
Partial correlation or multiple regression? They are closely related. If you want to predict the outcome from several variables at once, report a multiple regression.
What is a semi-partial correlation? It removes the control variable from only one of the two variables. It is reported the same way; name it clearly.
Can I control for more than one variable? Yes; list them all and subtract one df for each.