Pearson Correlation Calculator

What Is Pearson Correlation?

Pearson correlation coefficient ($r$) measures the strength and direction of the linear relationship between two continuous variables. It quantifies how closely data points follow a straight line.

Question answered: "How strongly are these two variables linearly related?"

Examples:

• "How strongly does study time relate to exam scores?"
• "Is there a relationship between height and weight?"
• "Do advertising spending and sales move together?"

The Correlation Coefficient

$r$ ranges from -1 to +1:

Understanding Direction

Positive correlation ($r > 0$): As X increases, Y tends to increase. Example: Height and weight

Negative correlation ($r < 0$): As X increases, Y tends to decrease. Example: Price and demand

Zero correlation ($r \approx 0$): No linear pattern. Note: Non-linear relationships may have $r \approx 0$!

The Formula

$r = \frac{\sum_{i=1}^{n}(x_i - \bar{x})(y_i - \bar{y})}{\sqrt{\sum_{i=1}^{n}(x_i - \bar{x})^2 \times \sum_{i=1}^{n}(y_i - \bar{y})^2}}$

Or equivalently:

$r = \frac{S_{xy}}{\sqrt{S_{xx} \times S_{yy}}}$

Where:

• $S_{xy}$ = sum of cross-products of deviations
• $S_{xx}$ = sum of squared deviations for X
• $S_{yy}$ = sum of squared deviations for Y

Example Calculation

Data: Study hours (X) and exam scores (Y)

Calculations:

• $\bar{x} = 4$, $\bar{y} = 74.4$
• $S_{xy} = 36$, $S_{xx} = 10$, $S_{yy} = 258.8$

$r = \frac{36}{\sqrt{10 \times 258.8}} = \frac{36}{50.87} = 0.708$

Strong positive correlation: more study hours associate with higher scores.

Coefficient of Determination ($r^2$)

$r^2 = \text{proportion of variance in Y explained by X}$

For $r = 0.708$: $r^2 = 0.50$

Interpretation: 50% of variation in exam scores is "explained" by study hours.

The remaining 50% is due to other factors.

Testing Significance

Is the correlation significantly different from zero?

$t = r \times \sqrt{\frac{n-2}{1-r^2}}$

With $df = n - 2$

For $r = 0.708$, $n = 5$:

$t = 0.708 \times \sqrt{\frac{3}{0.50}} = 0.708 \times 2.45 = 1.73$

P-value $\approx 0.18$ (not significant at $\alpha = 0.05$)

With only 5 points, even moderate correlation may not be significant.

Use correlation p-value calculator for precise values.

Sample Size and Significance

Small correlations become significant with large samples. Always report $r$ alongside p-value.

Correlation ≠ Causation

Critical warning: Correlation does not imply causation.

Possible explanations for correlation:

1. X causes Y
2. Y causes X
3. Third variable causes both
4. Coincidence

Example: Ice cream sales and drowning deaths are correlated. Reality: Hot weather causes both. Ice cream doesn't cause drowning.

When Pearson Correlation Fails

Non-Linear Relationships

A perfect parabolic relationship may have $r = 0$.

Always plot your data first!

Outliers

Single extreme points can dramatically change $r$.

Outliers may inflate or mask true correlation.

Restricted Range

If X has limited variability, correlation is attenuated.

Studying only honors students may hide the full correlation between study time and grades.

Heteroscedasticity

If Y's spread changes across X values, Pearson $r$ may be misleading.

Assumptions

Linearity

The relationship between X and Y is linear.

Check with scatterplot.

Continuous Variables

Both X and Y are measured on interval or ratio scales.

For ordinal data, use Spearman correlation.

Bivariate Normality

For significance testing, the joint distribution should be approximately normal.

No Significant Outliers

Outliers disproportionately influence $r$.

Homoscedasticity

The variance of Y is constant across levels of X.

Applications

Research and Science

Testing hypothesized relationships between variables.

"Is there evidence of a relationship between exposure and outcome?"

Finance

Measuring how assets move together.

Portfolio diversification relies on low or negative correlations.

Quality Control

Relating process parameters to output quality.

"How strongly does temperature affect product strength?"

Psychology

Validating measurement instruments.

Test-retest reliability, convergent validity.

Medicine

Relating biomarkers to outcomes.

"How well does this marker predict disease severity?"

Correlation Matrix

For multiple variables, compute all pairwise correlations:

Matrix is symmetric; diagonal is always 1.

Partial Correlation

Control for a third variable:

$r_{xy.z}$ = correlation between X and Y, holding Z constant.

Reveals whether X-Y relationship persists after accounting for Z.

Confidence Interval for r

Use Fisher's z-transformation:

$z' = \frac{1}{2} \ln\left(\frac{1+r}{1-r}\right)$

$SE$ of $z' = 1/\sqrt{n-3}$

Calculate CI in z-space, then transform back.

Comparing Correlations

To test if two correlations differ:

Convert both to $z'$ (Fisher transformation)

$z = \frac{z'_1 - z'_2}{\sqrt{\frac{1}{n_1-3} + \frac{1}{n_2-3}}}$

Reporting Results

APA format:

"There was a strong positive correlation between study hours and exam scores, r(3) = .71, p = .18. However, with only 5 participants, this correlation did not reach statistical significance."

For significant results:

"Hours of sleep was significantly correlated with reaction time, r(98) = -.45, p < .001, indicating that less sleep was associated with slower responses."

Historical Context

Karl Pearson developed this correlation coefficient in the 1890s, building on earlier work by Francis Galton.

Galton first used the concept of correlation to study heredity—how parent and offspring characteristics relate.

Pearson formalized the mathematics and gave us the formula still used today. His work established correlation as a fundamental tool for understanding relationships between variables.

Common Mistakes

Confusing Correlation with Causation

The most common error. Always consider alternative explanations.

Ignoring Non-Linearity

A scatterplot reveals patterns that $r$ misses.

Over-Interpreting Small Correlations

With large $n$, trivially small $r$ can be "significant."

Reporting $r^2$ as $r$

$r = 0.30$ is not the same as $r^2 = 0.30$.

Related Calculations

• Spearman Correlation — Rank-based correlation
• Correlation P-Value — Significance test
• R-Squared Calculator — Coefficient of determination
• Linear Regression — Predict Y from X
• Scatterplot Interpretation — Visual analysis

Summary

Pearson correlation coefficient ($r$) measures the linear relationship between two continuous variables, ranging from -1 (perfect negative) to +1 (perfect positive). Calculate using the formula involving cross-products of deviations, then test significance using a t-test with $df = n-2$. Always visualize data with a scatterplot first—$r$ only captures linear relationships. Report both $r$ and p-value, recognize that correlation does not imply causation, and be cautious with outliers. For ordinal data or non-linear monotonic relationships, use Spearman correlation instead.