How to Report Spearman's Correlation in APA Format

You ran the analysis, SPSS gave you a number, and now your supervisor wants the result reported cleanly in Chapter 4. Learning how to report Spearman's correlation in APA format is simpler than most students fear: two sentences, one set of numbers, and six formatting rules. This guide gives you the exact APA 7th template, the SPSS clicks to get the numbers, and a hand-worked example you can follow line by line.

What Is Spearman's Rank Correlation?

Spearman's rank correlation (Spearman's rho; the sample statistic is written r_s, and ρ is the Greek letter rho, used for the population value) measures the strength and direction of a monotonic relationship between two variables. Monotonic means the direction stays consistent: as one variable increases, the other tends to increase (a positive relationship) or tends to decrease (a negative one) — but the rate does not have to be constant, and the pattern does not have to be a straight line. Because it works on ranked data, it does not require your variables to be normally distributed or linearly related.

Like Pearson's r, Spearman's rho ranges from −1 to +1: +1 is a perfect positive monotonic relationship, −1 a perfect negative one, and 0 indicates no monotonic association between the ranks. Be careful with that last one: a strong curved pattern that changes direction (such as a U-shape) can hide behind a rho close to 0, which is why you always look at the scatterplot.

When to Use Spearman vs Pearson Correlation

Choose Spearman when:

One or both variables are ordinal — the classic example is single Likert-scale items (1 = strongly disagree … 5 = strongly agree).
The scatterplot shows a curved but consistently rising (or falling) pattern — monotonic, but not linear.
Outliers or extreme values are present that you cannot legitimately remove — because Spearman uses ranks, a single extreme score cannot dominate the result the way it can with Pearson.

A common oversimplification to avoid: "my data failed a normality test, so I must use Spearman." A failed normality test on its own is not a sufficient reason to switch. Pearson's r measures linear association, and it is the shape of the relationship (linear vs monotonic), the measurement level, and outliers that should drive your choice — let the scatterplot, not a single normality p-value, make the final call.

Choose Pearson when both variables are continuous, the scatterplot shows a roughly linear pattern, and there are no influential outliers. Strictly, it is Pearson's significance test — not the coefficient itself — that assumes the pair is approximately bivariate normal, and at moderate-to-large sample sizes it is fairly robust. If you want to describe how your data are distributed first, our guide on how to test normality in SPSS walks through it.

Before You Run It: The Assumptions Box

Spearman's rho has few assumptions, but they are real:

Paired observations — every case has a value on both variables.
At least ordinal measurement — both variables can be meaningfully ranked.
A monotonic relationship — checked on the scatterplot (Step 1 below).
Independent pairs — one pair per participant or unit; no repeated measures, and no clustered pairs (for example, several students from the same classroom treated as fully independent). Dependent pairs invalidate the p-value.
Tied values are allowed — SPSS assigns average ranks automatically (this matters for hand calculations; see the worked example).

How to Run Spearman Correlation in SPSS Step by Step

Check monotonicity first. Go to Graphs → Chart Builder → Scatter/Dot (or Graphs → Legacy Dialogs → Scatter/Dot → Simple Scatter), put one variable on the X axis and one on the Y axis. You want points that consistently rise or consistently fall — not a U-shape.
Click Analyze → Correlate → Bivariate. The Bivariate Correlations dialog opens. (This menu path has been stable across recent SPSS versions.)
Move your two variables into the Variables box.
Under Correlation Coefficients, untick Pearson and tick Spearman.
Under Test of Significance, keep Two-tailed (the default) unless you specified a direction in advance. Keep Flag significant correlations ticked so SPSS marks significant results with asterisks.
Click Options. Under Missing Values, the default is Exclude cases pairwise — each correlation uses every case that has valid values on that pair, which is why N can differ from pair to pair in a larger matrix. Choose Exclude cases listwise instead if your analysis plan requires the same cases throughout. You can also tick Means and standard deviations if you want descriptives in the same output. Click Continue.
Click OK. SPSS adds a block headed Nonparametric Correlations, containing a table titled Correlations — that table is your result. Note that the table is a matrix: every pair appears twice (once in each direction). Read one off-diagonal cell.

Prefer syntax? The same analysis, pasted from the dialog, is:

NONPAR CORR
  /VARIABLES=Overtime Stress
  /PRINT=SPEARMAN TWOTAIL.

(Replace the variable names with your own; check the pasted syntax against your SPSS version.)

How to Interpret the Spearman Correlation SPSS Output

The table gives you three numbers for each variable pair:

Correlation Coefficient (Spearman's rho) — the headline coefficient. Read the direction from its sign and the strength from the bands below.
Sig. (2-tailed) — the p-value. Report the exact value SPSS gives you. A value below your pre-set alpha (conventionally .05) is described as statistically significant — but treat the line as a convention, not a cliff edge: p = .049 and p = .051 represent almost identical evidence.
N — the number of cases actually used for that pair. Watch it: missing data can shrink it.

For strength, Cohen (1988) proposed the well-known guidelines .10 = small, .30 = medium, .50 = large for correlation coefficients (the same bands apply to negative values). Researchers commonly borrow these bands for Spearman's rho — treat them as rough guidance tied to your field, not as cutoffs or grades. Gignac and Szodorai (2016), in an analysis of 708 meta-analytically derived correlations, found that Cohen's bands are demanding for individual-differences research and suggested .10, .20, and .30 as relatively small, typical, and relatively large in that context — so in psychology and education, a rho of .30+ can be a perfectly respectable finding.

Two traps to avoid: a tiny rho can be "significant" with a huge sample, so always report the size and the p-value together; and correlation never proves causation — write "associated with", never "caused".

Worked Example: Computing Spearman's Rho by Hand

This is a completely hypothetical, illustrative example (not real study results). Suppose a researcher asks 10 employees about their weekly overtime hours and their stress scores (0–30):

Employee	Overtime (hours/week)	Stress score
1	7	18
2	5	10
3	8	15
4	4	9
5	6	14
6	9	20
7	3	8
8	10	22
9	6.5	6
10	8.5	12

Step 1 — rank each variable (1 = lowest). Overtime ranks: 6, 3, 7, 2, 4, 9, 1, 10, 5, 8. Stress ranks: 8, 4, 7, 3, 6, 9, 2, 10, 1, 5.

Step 2 — find d (rank difference) and d² for each pair: −2, −1, 0, −1, −2, 0, −1, 0, 4, 3 → squared: 4, 1, 0, 1, 4, 0, 1, 0, 16, 9. Sum of d² = 36.

Step 3 — apply the formula: ρ = 1 − (6 × Σd²) / (n × (n² − 1)) = 1 − (6 × 36) / (10 × 99) = 1 − 216/990 = 1 − 0.218 = .78.

Important — when this formula works. This shortcut formula is exact only when there are no tied ranks, which is true for this dataset. When values tie — and Likert-scale data are full of ties — give each tied case the average of the ranks they share, and compute Pearson's r on the ranks instead. That is what SPSS does internally, which is why a hand calculation with the shortcut formula can differ slightly from SPSS output on tied data.

Step 4 — interpret. ρ = .78 is a large positive effect by Cohen's bands: as overtime hours rise, stress scores generally rise. Notice Employee 9, though — 6.5 hours of overtime but the lowest stress score (rank difference of 4). That single break in the pattern is exactly why rho is .78 and not 1.00. With n = 10, the two-tailed critical value at α = .05 is about .65 (published tables give .648–.649) — our .78 exceeds it, so the result is statistically significant (p < .05). It does not reach the .01 critical value (.794), so from the table alone we can only claim p < .05.

Why SPSS shows a different p-value for the same data. If you type these 10 cases into SPSS, it prints Sig. (2-tailed) = .008 — not just "p < .05". SPSS calculates Spearman's significance using a t-approximation, which is slightly optimistic at very small samples like n = 10; an exact permutation test for these data gives p = .011. All three approaches agree on the conclusion at α = .05. In your own thesis, report the exact value your software gives you — and at very small sample sizes, treat that p-value as approximate.

A caution about small samples. With only n = 10, this estimate is imprecise: an approximate 95% confidence interval for a rho of .78 at this sample size runs from about .30 to .95 (Fisher's z approximation). A "large" rho from 10 cases should always be interpreted cautiously.

For reference, here is what the SPSS Correlations table contains for this hypothetical example (recreated from the values above — it is an illustration, not output from a real study):

		Overtime	Stress
Spearman's rho	Overtime — Correlation Coefficient	1.000	.782**
	Overtime — Sig. (2-tailed)	.	.008
	Overtime — N	10	10
	Stress — Correlation Coefficient	.782**	1.000
	Stress — Sig. (2-tailed)	.008	.
	Stress — N	10	10

** Correlation is significant at the 0.01 level (2-tailed), based on SPSS's t-approximation (see the note above).

How to Report Spearman Correlation in APA 7th Format

Here is the template (two sentences — the first names the test and variables, the second carries the result):

A Spearman's rank correlation was computed to assess the relationship between [Variable 1] and [Variable 2]. There was a [positive/negative] [strength word] correlation between the two variables, r_s(df) = [value], p = [value].

Filled in with our example, reported from the hand calculation:

A Spearman's rank correlation was computed to assess the relationship between weekly overtime hours and stress scores. There was a strong positive correlation between the two variables, r_s(8) = .78, p < .05.

And the same result as you would report it from SPSS output, using the exact Sig. value:

A Spearman's rank correlation was computed to assess the relationship between weekly overtime hours and stress scores. There was a strong positive correlation between the two variables, r_s(8) = .78, p = .008.

State your sample size and descriptive statistics in the surrounding text (for example, in your participants section or a descriptives table) — the df in the sentence already carries it, since df = N − 2.

The APA 7th formatting rules behind that sentence:

Italicise r and p. The s in r_s is a subscript identifier telling the reader the coefficient is Spearman's rho, not Pearson's r — APA sets identifier subscripts (and Greek letters such as ρ) in regular, non-italic type, so only the r itself is italic.
Degrees of freedom in parentheses right after r_s: df = N − 2 (here 10 − 2 = 8). Note: some journals instead ask for N in the parentheses — if your target journal's examples do that, follow the journal.
Round rho to two decimals and the p-value to three decimals.
Drop the leading zero for rho and p — write .78, not 0.78 — because neither value can exceed 1.
Never write p = .000. SPSS prints .000 for tiny values; report p < .001 instead.
Give the exact p-value when you have it (e.g., p = .008). Use p < .05 only when reporting from a critical-value table rather than software output.

One consistency tip: Cohen labels his bands small/medium/large, while narrative text often uses weak/moderate/strong for the same ranges. Either set is acceptable — pick the set your supervisor or journal uses and stay with it throughout your thesis.

Common Mistakes When Reporting Spearman Correlation

Writing r instead of r_s — the reader cannot tell which correlation you ran.
Reporting the coefficient without df and p — both belong in the sentence. (The df carries your N: N = df + 2. State N explicitly in your participants or descriptives section.)
Keeping the leading zero (0.78) or writing p = .000.
Calling .35 "weak" — under Cohen's bands .35 is already medium, and under Gignac and Szodorai's (2016) norms it is relatively large for individual-differences research. The bands are guidance that depends on your field, not grades.
Using the shortcut d² formula on tied data — with ties, average the ranks and correlate the ranks (or simply let SPSS do it).
Claiming causation — "overtime increased stress" goes beyond what a correlation can show.

Frequently Asked Questions

What is the difference between Pearson and Spearman correlation?

Pearson measures linear relationships between two continuous variables. Spearman measures monotonic relationships using ranks, so it works with ordinal data, monotonic-but-curved patterns, and data where extreme values would distort Pearson. Our Pearson correlation in SPSS guide covers the parametric side in the same step-by-step style.

Can I use Spearman correlation for Likert scale data?

Yes. Single Likert items are ordinal, which is Spearman's classic use case. Because Likert responses contain many tied values, let SPSS do the ranking — it assigns average ranks automatically. (For summed scale totals that behave approximately continuously and look linear on a scatterplot, Pearson is also a common choice.)

What does a negative Spearman rho mean?

As one variable's ranks rise, the other's ranks consistently fall — a steady inverse relationship, such as more sleep associated with lower stress rankings.

Do my data need to be normal for Spearman correlation?

No. Normality is not an assumption of Spearman's rho. You need paired, independent observations measured at the ordinal level or above, and a monotonic relationship between them.

Why does my SPSS p-value not match the critical-value table?

Because they are calculated differently. SPSS uses a t-approximation for Spearman's significance, which can differ slightly from exact table values at small sample sizes (in our worked example: SPSS p = .008, exact p = .011, table conclusion p < .05). Report the software's exact value, and treat it as approximate when your sample is very small.

Spearman or Kendall's tau-b — which should I use?

Both measure rank association. Kendall's tau-b is often preferred when your sample is very small or your data contain a large number of ties, because it handles ties explicitly and its values are more conservative. If your supervisor or journal expects Spearman, report Spearman — but know the alternative exists, and do not report both and pick the significant one.

Conclusion

Spearman's correlation is the right tool when your data are ordinal, when the relationship is monotonic but not linear, or when extreme values would distort a Pearson analysis — and reporting it in APA 7th comes down to one template: r_s(df) = value, p = value, with italics on r and p, two-decimal rho, and no leading zero. Run the Bivariate Correlations dialog, read rho for strength and Sig. (2-tailed) for the exact p-value, check that your pairs are independent and your ranks handled ties correctly, and your results section practically writes itself.

References

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates.

Gignac, G. E., & Szodorai, E. T. (2016). Effect size guidelines for individual differences researchers. Personality and Individual Differences, 102, 74–78. https://doi.org/10.1016/j.paid.2016.06.069

IBM. (n.d.). Bivariate correlations: Options. IBM SPSS Statistics documentation. https://www.ibm.com/docs/en/spss-statistics/saas?topic=correlations-bivariate-options

Need help with your thesis or SPSS analysis? WhatsApp Farzana at +92 326 2047345.

Struggling with SPSS for your thesis? Get the step-by-step guide SPSS Made Simple — 30-page ebook + Excel templates pack. Buy it here: https://whop.com/researchmadesimple/spss-made-simple-a-step-by-step-guide-for-psychology-social-science-students/

Search This Blog

Thesis and SPSS Help by Farzana