> r1 <- round(rnorm(20, 10, 4))
> r2 <- round(r1 + 10 + rnorm(20, 0, 2))
> r3 <- round(r1 + 20 + rnorm(20, 0, 2))
> icc(cbind(r1, r2, r3), "twoway")
Single Score Intraclass Correlation

   Model: twoway
   Type : consistency

   Subjects = 20
     Raters = 3
   ICC(C,1) = 0.917

F-Test, H0: r0 = 0 ; H1: r0 > 0
   F(19,38) = 34.1 , p = 3.43e-18

95%-Confidence Interval for ICC Population Values:
  0.835 < ICC < 0.963
> icc(cbind(r1, r2, r3), "twoway", "agreement")
Single Score Intraclass Correlation

   Model: twoway
   Type : agreement

   Subjects = 20
     Raters = 3
   ICC(A,1) = 0.2

F-Test, H0: r0 = 0 ; H1: r0 > 0
F(19,2.22) = 34.1 , p = 0.0211

95%-Confidence Interval for ICC Population Values:
  0.001 < ICC < 0.526

> r1
[1]  9  7 16 12  7 16 14 14 18 17  3  8  7  6 14 10  6 18 19 14
> r2
[1] 19 19 24 20 15 23 29 24 27 26 12 17 15 19 24 21 14 32 29 25
> r3
[1] 30 23 35 32 27 35 36 35 36 39 25 28 27 24 32 30 25 39 40 35

这个consistency是很高的，0.917，但是agreement比较低，只有0.2，对于agreement很好理解，我是学医的，agreement是非常重要的，否则医学实践就无法进行。但是consistency是什么，这三组变化这么大的数据，consistency却很高，是不是有问题？