Other Correlation Approaches
- PEARSON: great for normally-distributed metric data (also frequently referred to as interval data); sensitive to outliers; significance test distorted by non-normality.
- KENDALL’S TAU – appropriate for ordinal data such as ordered categories with unknown distances between them (e.g., disagree, neutral, agree); not sensitive to outliers; significance test not sensitive to distributional anomalies.
- SPEARMAN’S RANK-ORDER CORRELATION – as the name suggests, appropriate for rank data (distance between ranks is usually unknown); not sensitive to outliers; significance test not sensitive to distributional anomalies.