|
The Spearman rank correlation coefficent (above) uses ranks
and is not dependent upon the distribution of either variable. Both variables are ranked from 1 to n, the differences in ranks are calculated, squared, and then summed.
Bootstrap PLUS .... the computer can be used to calculate the p-
value for this rank correlation coefficient. Repeated random samples of data pairs (X and Y values) are taken after one of the variables has been 'shuffled'. The rank correlation coefficent is calculated and compared to the original value. This is repeated many times (e.g. 1000 times). The number of times, out of 1000, that the calculated rank correlation coefficient exceeds the orginal value is used to estimate of the p-value. |