1

For Wilcoxon test, I have p=0.000. I also did the Kendall's test to test the correlation, and I have Tb=0.366. I don't get it, if there is a significant difference, why there is a moderate agreement?

Parastoo
  • 11
  • 1
    Kendall's test does not measure agreement; it measures correlation. Consider $y = x +\ $ an enormous constant by comparison. The correlation is perfect but the agreement is lousy – Nick Cox Oct 27 '16 at 16:26
  • Concordance correlation: search the forum for mentions. – Nick Cox Oct 27 '16 at 16:44
  • @NickCox Is there a missing symbol in your first comment? – Glen_b Oct 28 '16 at 02:06
  • No; suppose $y$ and $x$ differ only by an enormous constant, enormous being relative to the scale of either variable. Then any kind of correlation between $y$ and $x$ is 1, but they don't agree. I take it that perfect agreement means equality. – Nick Cox Oct 28 '16 at 08:32

0 Answers0