Rank Order Correlation

• The correlation coefficient(r) is based on the assumption that data is normally distributed.
• If the data are skewed, instead of r we use a rank order correlation rs (also called spearman coefficient) as follows.

• Where d=difference between the ranks of each pair
• N=the number of pair of observations
• The spearman coefficient will assume any value between -1.00 to + 1.00 inclusive.
• -1.00 indicates a perfect negative correlation
• +1.00 indicates a perfect positive correlation
• The ranking order has to be done for each pair of data
• The highest data value will be lowest rank or the other way.
• The lowest data value will be highest rank or the other way.