Correlation analysis summarizes the direction and consistency with which two variables move together. It is an exploratory description of association, not evidence that changing one variable would change the other.

The central question is: “When one variable changes, does the other tend to move in a consistent direction, how strong is that pattern, and how uncertain is the estimate?” Defining it before opening a menu prevents a page of statistics from replacing an explanation.

The purpose is the direction and consistency of co-movement without losing the shape of the relationship. Interpret the coefficient, 95% interval, sample size, scatterplot, and influential observations together, because outliers, restricted range, and mixed subgroups can create or reverse a coefficient.

Pearson r measures linear association, Spearman rho measures monotonic rank association, and Kendall tau is based on concordant pairs. Sign gives direction and magnitude gives sample association, but neither describes a curved shape or causal mechanism.

No single statistic is sufficient. Use scatterplots to find curvature, clusters, restricted range, and influential points. Large correlation matrices create chance findings; use prespecified questions, confidence intervals, and multiplicity control.

The purpose is the direction and consistency of co-movement without losing the shape of the relationship. Interpret the coefficient, 95% interval, sample size, scatterplot, and influential observations together, because outliers, restricted range, and mixed subgroups can create or reverse a coefficient.

Pearson r measures linear association, Spearman rho measures monotonic rank association, and Kendall tau is based on concordant pairs. Sign gives direction and magnitude gives sample association, but neither describes a curved shape or causal mechanism.

No single statistic is sufficient. Use scatterplots to find curvature, clusters, restricted range, and influential points. Large correlation matrices create chance findings; use prespecified questions, confidence intervals, and multiplicity control.

The purpose is the direction and consistency of co-movement without losing the shape of the relationship. Interpret the coefficient, 95% interval, sample size, scatterplot, and influential observations together, because outliers, restricted range, and mixed subgroups can create or reverse a coefficient.

Pearson r measures linear association, Spearman rho measures monotonic rank association, and Kendall tau is based on concordant pairs. Sign gives direction and magnitude gives sample association, but neither describes a curved shape or causal mechanism.

No single statistic is sufficient. Use scatterplots to find curvature, clusters, restricted range, and influential points. Large correlation matrices create chance findings; use prespecified questions, confidence intervals, and multiplicity control.

The purpose is the direction and consistency of co-movement without losing the shape of the relationship. Interpret the coefficient, 95% interval, sample size, scatterplot, and influential observations together, because outliers, restricted range, and mixed subgroups can create or reverse a coefficient.

Pearson r measures linear association, Spearman rho measures monotonic rank association, and Kendall tau is based on concordant pairs. Sign gives direction and magnitude gives sample association, but neither describes a curved shape or causal mechanism.

No single statistic is sufficient. Use scatterplots to find curvature, clusters, restricted range, and influential points. Large correlation matrices create chance findings; use prespecified questions, confidence intervals, and multiplicity control.

The purpose is the direction and consistency of co-movement without losing the shape of the relationship. Interpret the coefficient, 95% interval, sample size, scatterplot, and influential observations together, because outliers, restricted range, and mixed subgroups can create or reverse a coefficient.

A defensible workflow is Check variable types → Plot the relationship → Choose the coefficient → Estimate with interval → Stress-test the pattern. Record the variable definitions, exclusions, transformations, denominators, and decision rules. Reproducibility begins with these choices, not with the final number.

Begin by auditing type, unit, coding, missingness, and Pearson, Spearman, or Kendall together with intervals and multiplicity control. Defaults are only starting points; document every denominator, exclusion, transformation, and decision rule needed to reproduce the result.

A correlation of r=.62 between study time and score may disappear if it is created by one influential case. A scatterplot should show whether the trend spans the observed range and whether grade-level subgroups tell a different story.

Report sample size and missingness alongside the primary estimate, comparison standard, uncertainty, and visible pattern. A descriptive result becomes inferential or causal only when the design supports that claim.

Treat the worked number as the beginning of interpretation. Read the coefficient, 95% interval, sample size, scatterplot, and influential observations together; then state what the sample supports without turning a descriptive pattern into a population or causal claim.

Common uses include relationships among survey scales, price–demand exploration, clinical marker associations, feature redundancy screening. Before operational use, define how the same quantity will be recomputed for new data, how subgroup differences will be monitored, and what action the result is meant to support.

The purpose is the direction and consistency of co-movement without losing the shape of the relationship. Interpret the coefficient, 95% interval, sample size, scatterplot, and influential observations together, because outliers, restricted range, and mixed subgroups can create or reverse a coefficient.

Use scatterplots to find curvature, clusters, restricted range, and influential points. Large correlation matrices create chance findings; use prespecified questions, confidence intervals, and multiplicity control.

Strengths include Compact direction and magnitude on a common scale; Useful starting point for regression and feature review; Rank methods extend to ordinal and non-normal data. Important limitations are Correlation is not causation; A strong nonlinear relationship can have r near zero; Outliers, range restriction, and mixed subgroups can dominate.

Diagnostics are part of the result because outliers, restricted range, and mixed subgroups can create or reverse a coefficient. Compare reasonable alternatives and inspect the observations that drive the summary before treating one output as stable.

A correlation coefficient is a compressed scatterplot. Report the coefficient, confidence interval, sample size, and visible shape together. Recommended sequence: Check variable types → Plot the relationship → Choose the coefficient → Estimate with interval → Stress-test the pattern.

A complete report links the question, analysis population, Pearson, Spearman, or Kendall together with intervals and multiplicity control, the primary result, and the diagnostic evidence. A reader should be able to reconstruct the same quantity from those decisions.