Start Here
Not sure which test to use? The Correlation Wizard asks a few questions and gives you a direct recommendation.
Open the Correlation Wizard →Correlation Analysis Hub
Quantifying the strength and direction of a relationship between two variables.
Related Calculators
Related Articles
Medical Research Examples
Pearson & Spearman Correlation: A nephrologist studies serum creatinine vs. 24-hour urinary protein in 45 CKD patients: Pearson r=0.74, Spearman ρ=0.79 (both p<0.001); R²=0.55 -- serum creatinine explains 55% of the variance in proteinuria.
Common Mistakes
- Using Pearson r on skewed or ordinal data (e.g. pain scores, Likert scales) instead of Spearman.
- Interpreting a significant correlation as proof of causation.
- Ignoring outliers, which can dramatically inflate or deflate Pearson's r.
- Reporting r without r² or a confidence interval, leaving readers unable to judge practical importance.
- Assuming a weak r (e.g. 0.2) is unimportant without considering sample size and context -- a small r can still be clinically meaningful in a large study.
- Not checking for a non-linear (e.g. U-shaped) relationship, which can produce a near-zero Pearson r despite a strong underlying association.
Reporting Recommendations
Journals expect more than a p-value. For each test, report the effect estimate and its 95% confidence interval alongside the p-value:
Correlation Coefficient (Pearson / Spearman)r or ρ, 95% CI, p-value, sample size n, r² (variance explained)
Recommended Learning Order
- Confidence Interval & P-Value fundamentals — understand the building blocks of inferential statistics
- Normality Testing — decide whether Pearson or Spearman is valid for your data
- Correlation Analysis (Pearson/Spearman) — quantify the strength of a relationship between two variables
- Simple Linear Regression — natural next step once a linear relationship is confirmed
- Multiple / Logistic Regression — extend to multiple predictors or binary outcomes
Frequently Compared Tests
Pearson vs. Spearman vs. Kendall's Tau
Kendall's Tau has no dedicated calculator on StatClinic yet -- treated here as a concept, not a live tool.
Frequently Asked Questions
When do I use Pearson vs. Spearman?
Pearson requires both variables to be continuous and approximately normally distributed. Spearman is for ordinal, skewed, or outlier-prone data.
What does Kendall's Tau add over Spearman?
Tau is more robust with small samples and many tied ranks, and its value is interpreted directly as a probability of concordance. StatClinic doesn't yet have a dedicated Tau calculator -- see Frequently Compared Tests below.
What sample size do I need for a correlation study?
Use the Sample Size Calculator. Correlation studies typically need larger samples than group-comparison studies to estimate r precisely.
Does a significant correlation mean one variable causes the other?
No. Correlation only establishes association. Causal inference requires a designed experiment or a method that adjusts for confounding, such as multiple regression.
My variables aren't linearly related but still seem connected -- what now?
Check a scatterplot. If the relationship is monotonic but curved, Spearman's ρ will still capture it even though Pearson's r will underestimate the strength.
How do I report a correlation result in a paper?
Report r or ρ, its 95% confidence interval, the exact p-value, sample size n, and r² (proportion of variance explained).