SPSS command: SPSS CORRELATIONS
Pearson, Spearman and Kendall correlations in your browser
A correlation coefficient measures how closely two variables move together, from −1 to +1.
Free · No account · Files stay on your device · Updated 18 October 2026
What it tells you
Crosstabs builds SPSS-style CORRELATIONS and NONPAR CORR matrices on your device: Pearson r for linear association, Spearman's rho and Kendall's tau-b for monotonic association, each with significance, N and optional flags.
- The direction and strength of association for every pair of selected variables.
- Whether each coefficient differs from zero, with one- or two-tailed significance and * / ** flags at the .05 and .01 levels.
- A 95% confidence interval for Pearson r from Fisher's z transformation, showing how precisely the correlation is estimated.
- Whether a rank coefficient tells a different story from Pearson, a sign of outliers or a curved but monotonic relationship.
When to use it
Use it when
- Screening many survey items or KPIs at once to see which move together.
- Two scale variables with a roughly linear relationship: Pearson r with its confidence interval.
- Ordinal ratings, skewed variables or data with outliers: Spearman's rho or Kendall's tau-b.
Use something else when
- A third variable may drive both: use partial correlation to control for it.
- You want to predict one variable from several others: use linear regression.
- Both variables are nominal categories: use Crosstabs with chi-square and Cramér's V instead.
How to run it in Crosstabs
- 01
Choose Bivariate correlations
In Statistics mode open Correlate → Bivariate correlations (CORRELATIONS).
- 02
Tick the variables
Under Variables tick at least two numeric variables. Every pair appears in the matrix.
- 03
Pick coefficients and tails
Under Correlation Coefficients tick Pearson, Kendall's tau-b and/or Spearman, then choose Two-tailed or One-tailed. Optionally tick Flag significant correlations and 95% confidence interval for Pearson (Fisher z).
- 04
Choose missing-value handling
Exclude cases pairwise uses every case valid on both variables of a pair; Exclude cases listwise keeps one common set of complete cases for the whole matrix.
- 05
Run correlations
Click Run correlations. If a workspace weight is set, declare it first: a frequency weight is applied, a sampling weight gives weighted Pearson coefficients only.
Worked example
Computed by the Crosstabs engine from the synthetic data below when this page was built — not typed in.
Data
40 synthetic employees with years of tenure (right-skewed), an engagement score and days absent last year. Engagement rises slightly with the log of tenure and absence falls as engagement rises; all three are drawn with a fixed seed.
| Item | tenure | engagement | absence |
|---|---|---|---|
| tenure — Pearson Correlation | 1.000 | 0.166 | 0.013 |
| tenure — Sig. (2-tailed) | .306 | .938 | |
| tenure — 95% CI | −0.153, 0.454 | −0.300, 0.323 | |
| engagement — Pearson Correlation | 0.166 | 1.000 | −0.367* |
| engagement — Sig. (2-tailed) | .306 | .020 | |
| engagement — 95% CI | −0.153, 0.454 | −0.609, −0.063 | |
| absence — Pearson Correlation | 0.013 | −0.367* | 1.000 |
| absence — Sig. (2-tailed) | .938 | .020 | |
| absence — 95% CI | −0.300, 0.323 | −0.609, −0.063 |
N = 40 for every pair. **. Significant at the 0.01 level (2-tailed). *. Significant at the 0.05 level (2-tailed). CIs use Fisher's z.
| Item | tenure | engagement | absence |
|---|---|---|---|
| tenure — Spearman's rho | 1.000 | 0.263 | −0.067 |
| tenure — Sig. (2-tailed) | .102 | .683 | |
| engagement — Spearman's rho | 0.263 | 1.000 | −0.350* |
| engagement — Sig. (2-tailed) | .102 | .027 | |
| absence — Spearman's rho | −0.067 | −0.350* | 1.000 |
| absence — Sig. (2-tailed) | .683 | .027 |
How to read it
Engagement and absence move in opposite directions: Pearson r = −0.37 (95% CI −0.61 to −0.06, p = .020) and Spearman's rho = −0.35. For tenure, whose distribution is skewed, Pearson gives r = 0.17 and Spearman rho = 0.26; when the two disagree, the rank-based coefficient is the safer description of a monotonic but non-linear association.
Output tables explained
- Correlations
- The Pearson matrix: for each pair, Pearson Correlation, Sig. (2-tailed or 1-tailed), N and, if requested, the 95% CI (lower, upper). Footnotes explain the flags and the missing-value rule.
- Nonparametric Correlations
- One matrix for Kendall's tau_b and one for Spearman's rho, each with significance and N, laid out like the Pearson table.
Assumptions and notes
- Pearson r captures linear association only; a strong curved relationship can give r near zero.
- Pearson significance assumes cases are independent and the pair is roughly bivariate normal; rank coefficients need only ordinal data.
- With pairwise deletion each coefficient may rest on a different N, so compare coefficients with care.
- A frequency weight is used as given for Pearson (N = sum of weights); Spearman and Kendall replicate each case by its weight rounded to the nearest integer.
How the results are validated
The correlation engine is tested against independent reference values computed with SciPy 1.18.1, statsmodels 0.15.0, pingouin 0.7.0 and scikit-posthocs 0.17.1: Pearson, Spearman and Kendall tau-b coefficients with two- and one-tailed p-values match scipy.stats under both pairwise and listwise deletion, as do the Pearson Fisher-z intervals; integer frequency weights match SciPy run on replicated cases, and fractional weights for Pearson match statsmodels DescrStatsW with df = W − 2. Tolerance is 1e-9 (relative to the larger of 1 and the reference value). Validation evidence
Questions
Pearson, Spearman or Kendall: which should I use?
Use Pearson for roughly linear relationships between scale variables. Use Spearman for ordinal data or monotonic but curved relationships. Kendall's tau-b handles many ties well and is often preferred for small samples of ratings.
How is the confidence interval for r calculated?
With Fisher's z transformation: atanh(r) ± 1.96/√(N − 3), transformed back with tanh. No bias adjustment is applied, and intervals are offered for Pearson only.
Should I choose pairwise or listwise deletion?
Pairwise keeps more data per coefficient; listwise guarantees that every coefficient comes from the same cases, which matters if the matrix feeds another analysis.
When is a one-tailed test appropriate?
Only when the direction was predicted before looking at the data. The one-tailed p is half the two-tailed p in the predicted direction.
Does a significant correlation mean one variable causes the other?
No. It shows the variables move together in this data. A shared driver, reverse causation or selection can all produce the same coefficient.
Related procedures
- Partial CorrelationPARTIAL CORR
- Linear RegressionREGRESSION
- Bayesian Pearson CorrelationBAYES CORRELATION
- DescriptivesDESCRIPTIVES
- Crosstabs StatisticsCROSSTABS /STATISTICS
- Free SPSS alternative: full coverage list
Run Bivariate Correlations on your own data
Open a .sav, CSV or Excel file. Everything runs in your browser; the file never leaves your device.
SPSS is a trademark of IBM Corp. Crosstabs is not affiliated with or endorsed by IBM.