Quantitative/ Analysis
Correlation
Measure association without claiming agreement
Correlation quantifies association between variables. Pearson’s coefficient describes linear association in numeric values; Spearman and Kendall use ranks to describe monotonic association. A coefficient must be interpreted with its scatterplot, range and uncertainty. Repeated observations and clustering require methods that separate within-unit and between-unit relationships.
WHEN IT FITS
Use when the question concerns strength and direction of association between two measured variables. Inspect the shape of the relationship and decide whether linear values or ordinal ranks are meaningful.
Strengths
- Compact scale-free description of association
- Rank measures suit monotonic relationships and ordered data
Limitations
- Outliers and restricted ranges can distort results
- Pooling groups can conceal or reverse within-group patterns
Know the boundary
Correlation is neither causation nor agreement between measurement instruments.
USED ACROSS
Banking & financeBusiness & MBAPsychologyComputer scienceEducation