comparing dependent Spearman correlations

[Dave Airey]

Elsewhere I noted it would be nice to have a Stata command to compare dependent Spearman correlations. The first link below is an R package for comparing Pearson correlations that is pretty comprehensive. The second link is for theory and SAS macros comparing Spearman correlations with and without replication.

https://journals.plos.org/plosone/ar...l.pone.0121945

https://www.omicsonline.org/open-acc....php?aid=54592

I have since come across literature that suggests bootstrapping the difference in two dependent Spearman correlations is a reasonable thing to do. This is easy:

program spcordiff, rclass
    version 15.1
    args x y z
    tempname sp_xz sp_yz
    spearman `x' `z'
    scalar `sp_xz' = r(rho)
    spearman `y' `z'
    scalar `sp_yz' = r(rho)
    return scalar diff = `sp_xz'-`sp_yz'
end

bootstrap diff = r(diff), reps(1000): spcordiff x y z