What is the difference between Pearson, Spearman, and Kendall correlations?

Pearson correlation measures linear relationships and assumes normally distributed data. Spearman rank correlation measures monotonic relationships (not necessarily linear) and works with ordinal data. Kendall tau-b is more robust to outliers and ties, making it suitable for small samples or data with many tied values.

When should I use Pearson correlation?

Use Pearson correlation when you have continuous data that follows a normal distribution and you want to measure the strength of a linear relationship. It's ideal for interval or ratio scale data like height, weight, temperature, or test scores.

What is Fisher's z-transformation and why is it used?

Fisher's z-transformation converts the correlation coefficient (r) to a z-score, which follows a normal distribution. This allows us to calculate accurate confidence intervals for the correlation coefficient, which is essential for statistical inference.

How do I interpret a correlation coefficient?

Correlation coefficients range from -1 to +1. Values near 0 indicate weak or no correlation. Positive values indicate variables increase together, while negative values indicate one increases as the other decreases. Generally: 0.0-0.3 is weak, 0.3-0.7 is moderate, and 0.7-1.0 is strong correlation.

Correlation Calculator

Compute Pearson, Spearman, and Kendall correlations for a single pair or a matrix. Add confidence intervals, visualize scatter and heatmaps, and export results.

📚 Learn How to Use This Calculator

New to correlation analysis? Check out our comprehensive guide to understand the concepts and learn how to interpret your results.

1) Select Mode

Choose between single pair or correlation matrix analysis

Compare two variables (X and Y) with detailed statistics and scatter plots

2) Enter Data

Enter X and Y data manually or select CSV columns

X Variable

Y Variable

3) Options

Configure analysis settings

Confidence Level

For Pearson confidence intervals

4) Results

Correlation coefficients, p-values, and visualizations

Enter your data and click "Compute Correlation" to see results

Next steps

Notes & Assumptions

• Pearson: Assumes linearity, homoscedasticity, and approximate normality. Best for continuous data.
• Spearman: Measures monotonic relationships using ranks. More robust to outliers and non-normal data.
• Kendall: Ordinal association with better handling of tied values. More conservative than Spearman.
• Confidence Intervals: Fisher-z transformation for Pearson; bootstrap recommended for Spearman/Kendall (especially in matrix mode).
• Correlation ≠ Causation: Correlation only measures association, not causal relationships.
• Missing values are handled using pairwise deletion by default

Learn More

📘 Correlation analysis guide · 📘 Fisher-z confidence intervals · 📘 Correlation vs causation

Related calculators: Regression ·Descriptive ·Hypothesis Tests