在大样本时，即当n一→α时，有Z = rs Vn- 1 → N(0, 1)在X或Y有结时,令u1,u2，...和V1,V2,…分别代表X和Y的各个结的观测值数目记U= (ug - uj), V= (u -vi)调整过的Spearman统计量为n(n2 - 1) - 6Z;(Ri - Si)2 - 6(U + V)rs =V[n(n2 - 1) - Un(n2 - 1) - V)
例：儿童死亡率和母亲死亡率X56113 17 100 31 360 880110 32 54 78 110 1600 230Y438126839262 5 112 17 106 1461345821X51030055 51055026017015562230760130480143Y837468422733529412359321108问：X和Y是否有相关性？
例： 问： 是否有相关性？
a8oO00o00200o50000oOgooo08oO0088?008o8OCC00080000O4o00oo000008m0000S6%%o(K)601C000Yo0o8oo808.8800o8o030000o800880000o0008000%00oO8000.008%Co8RC000020o008CO000008800园8600008Co000oO1-I02354co7500100015002000log(x)x线性相关性：人用log变换后数据更合适
线性相关性：用log变换后数据更合适
Ri，Si及di的数值78Ri41524 29 13 1.5 16.50 910.00 14.00 16.50 30 20.50S;82741025 28 9 2.5 21.00 20 6.50 11.50 18.00 30 17.009di1641441 2.251 16 1.0 20.25 121 12.256.250 12.25619R;5.001223 11.00261.50 27 18 25 2220.50 28 3.001S;14166.5024 11.50265.00 23 13 22 1519.00 29 2.50252516d;491. 02.250.252.2510.2512.25 16.25 9
的数值
ex6.example1=function((d=read.table("D:/Nonpara/DM1.txt")x=d[,2];y=d[,1]rx=rank(x);ry=rank(y)rsd=rbind(rx,ry,(rx-ry)^2)list(rsd,cor.test(x,y,meth=c("pearson""spearman"))))> ex6.example1()4829715241330 20.50rx1.516.50910.0014.0016.50825428927102.521.0030 17.00ry206.5011.5018.00411691441161.020.25012.25121 12.25 6.25 2.25272522191823 11.00261.5065.001220.50283.00rx2322151413265.00124 11.506.501619.002992.50ry925491625251 0.25012.252.25 16 2.250.251
ex6.example1=function(){ d=read.table("D:/Nonpara/DM1.txt") x=d[,2];y=d[,1] rx=rank(x);ry=rank(y) rsd=rbind(rx,ry,(rx-ry)^2) list(rsd,cor.test(x,y,meth=c("pearson", "spearman")))} > ex6.example1() rx 4 7 15 8 24 29 13 1.5 16.50 9 10.00 14.00 16.50 30 20.50 ry 8 4 27 10 25 28 9 2.5 21.00 20 6.50 11.50 18.00 30 17.00 16 9 144 4 1 1 16 1.0 20.25 121 12.25 6.25 2.25 0 12.25 rx 23 11.00 26 1.50 27 18 25 22 19 6 5.00 12 20.50 28 3.00 ry 24 11.50 26 5.00 23 13 22 15 14 1 6.50 16 19.00 29 2.50 1 0.25 0 12.25 16 25 9 49 25 25 2.25 16 2.25 1 0.25