Skip to content
Back to search results

Pearson Correlation

Introduced by
Karl Pearson
Year
1895
Type
Parametric correlation
Font size

What does it do?

Pearson correlation measures the direction and strength of linear association between two variables.

It is their covariance divided by the product of their standard deviations.

The intuition behind the method

The coefficient compares whether centred values move together, after removing the influence of their measurement units.

When to use it

Use when linear association answers the research question and paired measurements are meaningful. Inspect the joint relationship and influential observations.

Limitations and considerations

  • Nonlinear relationships can be strong even when the coefficient is near zero.
  • Constant inputs make the coefficient undefined. Influential observations can distort its interpretation.
  • Correlation alone does not establish causality.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Provides a scale-free description of linear association.

Applications

  • Describing paired linear relationships and exploring feature associations.

How it works

How it works

Calculate the coefficient from paired observations with nonzero variation.

Separate calculation from inference. The conventional null test relies on its distributional assumptions; resampling methods have their own requirements.

1
Formula
H_0:\ \rho = 0 \quad H_1:\ \rho \neq 0
2
Formula
r = \dfrac{\sum_{i=1}^{n}(x_i - \bar{x})(y_i - \bar{y})}{\sqrt{\sum_{i=1}^{n}(x_i - \bar{x})^2\;\sum_{i=1}^{n}(y_i - \bar{y})^2}}
3
Formula
t = r\sqrt{\dfrac{n-2}{1-r^2}},\quad H_0: \rho=0

Common pitfalls

  • Requiring normality simply to calculate the coefficient.
  • Using a universal minimum sample size or significance threshold in place of a study-specific inferential plan.
  • Reporting association as a causal effect.

Related methods

Sources

  1. Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum Associates.

    Open DOI reference
    Copied.
  2. Field, A. (2013). Discovering Statistics Using IBM SPSS Statistics (4th ed.). SAGE.

    ISBN: 978-1446249185
    Copied.
  3. SciPy pearsonr documentation

    Open publication reference
    Copied.