I am trying to choose between the two correlation tests. I am trying to see if a trait is correlated with the area of a habitat patch for a number of different habitat types. The data is non-parametric and cannot be made normal through transformation, so the choice is between Spearman's or Kendall's
As it is comparing areas numerous data points will have the same area value which means there will be lots of ties in the data. I understand that Kendall's tau handles ties far better than Spearman's does.
However, some of the data sets are very large including one that is approximately 500,000 data points, and I understand that the cost of Kendall's is O(N^2) so it could potentially be very computationally demanding.
How do I make the choice between the two correlation statistics in this case?
