How to Report a Spearman Correlation in APA 7

Spearman is the non-parametric counterpart to Pearson and turns up in almost any paper with ordinal data or skewed scales. It is also one of the worst-written statistics: wrong symbol, no degrees of freedom, and no explanation of why Pearson was not used. Here is how to report it properly in APA 7. The parametric version is in how to report a correlation in APA 7.

The correct symbol

APA 7 accepts two notations: r_s (the more common one in psychology) and ρ (Greek rho). Pick one and keep it throughout the manuscript. What does not work is a bare r, because readers will take it as Pearson and your analysis is then read as something it is not.

The Latin r is set in italics, subscript included. Greek ρ, like every Greek letter in APA, is set upright. It is the same rule that keeps the β of a regression out of italics.

And since the coefficient is bounded between −1 and 1, it never takes a leading zero: write r_s = .42, not 0.42. The same applies to the p value.

Example in the text

"The relationship between educational level and job satisfaction was positive and moderate, r_s(98) = .42, p < .001. Spearman’s correlation was used because educational level was measured on a five-category ordinal scale."

Degrees of freedom are N − 2. With 100 participants, r_s(98). Many journals accept giving N directly instead — r_s = .42, N = 100, p < .001 — and that is equally correct as long as sample size is clear. What is not acceptable is a bare coefficient with neither.

Always write down why you used Spearman. One sentence is enough: ordinal variables, a monotonic but non-linear relationship, or outliers distorting Pearson’s r. An unjustified Spearman coefficient looks like what it sometimes is: a Pearson that came out non-significant.

Already drafted it? Paste your Results section into Review one section and you get the first objection a reviewer would raise, quoting the exact sentence from your text. Free, no sign-up.

When Spearman and when Pearson

Use Spearman if at least one variable is ordinal (a ranking, a Likert scale treated as order, a clinical stage), if the relationship is monotonic but clearly curved, or if influential outliers are present. Because it works on ranks, it is robust to all three.

Use Pearson if both variables are continuous, the relationship is roughly linear and no outlier dominates the coefficient. It is more powerful when its assumptions hold, so do not abandon it without a reason.

A third option almost nobody considers: Kendall’s tau. It is preferable to Spearman with small samples or many ties, and it interprets more cleanly as a probability of concordance. Report it the same way, with τ set upright because it is Greek.

What you must not do is compute both and keep whichever comes out significant. That decision belongs before you look at the results, based on the distribution and the type of variable, and it is described in the analysis section.

Effect size, intervals and matrices

r_s is already an effect size: no second one is needed. Conventional benchmarks put small around .10, medium around .30 and large around .50, but it is better interpreted against what is typical in your field than against a generic table.

APA 7 recommends adding the 95 % confidence interval. For Spearman it is obtained by bootstrap or by applying Fisher’s transformation to the ranks, and written like this: r_s = .42, 95 % CI [.24, .57], p < .001. How to build them is in reporting confidence intervals and bootstrap.

With several variables, use a correlation matrix rather than running text: variables as rows and columns, coefficients in the lower triangle, and a note explaining the asterisks. If you mix Spearman and Pearson in the same table, say so in the note and mark which is which — leaving it ambiguous is a classic flag. The APA table generator builds it for you.

Frequently asked questions

Is it written rs or rho for Spearman?

Both are correct in APA 7. rs, with the r in italics and a subscript s, is the more common form in psychology; ρ (Greek rho) is set upright because Greek letters are never italicized in APA. What matters is picking one and using it consistently, and never writing a bare r, which reads as Pearson.

Does Spearman’s correlation take degrees of freedom?

Yes, and they are N − 2, exactly as in Pearson: with 100 participants you write rs(98) = .42. Giving N explicitly instead of degrees of freedom is also acceptable. What is not acceptable is a coefficient with no reference to sample size at all.

When should I use Spearman instead of Pearson?

When at least one variable is ordinal, when the relationship is monotonic but not linear, or when outliers distort Pearson’s coefficient. Because Spearman works on ranks rather than raw values, it is robust in all three cases. The decision is made before seeing the results and explained in the analysis section.

Do I need an effect size in addition to rs?

No. The coefficient is already an effect size and adding another is redundant. What is worth adding, because APA 7 recommends it, is its 95 % confidence interval, obtained by bootstrap or via Fisher’s transformation.

Does rs take a zero before the decimal point?

No. Only statistics that can exceed 1 in absolute value take a leading zero. Since a correlation coefficient is bounded between −1 and 1, write rs = .42 and never 0.42. The same rule applies to p values and to R² in a regression.

Can I put Spearman and Pearson in the same table?

Yes, and it is common when some variables are continuous and others ordinal, but you must say so in the table note and make clear which coefficient each cell holds. The usual solution is Pearson in one triangle and Spearman in the other, explained in the note. Leaving it ambiguous is flagged in review.

With the right symbol, the degrees of freedom and one sentence justifying the choice, your Spearman correlation is properly reported. If you are unsure between Spearman, Pearson and Kendall for your data, we can look at it in my statistical consulting.