MACRO KENPAR for Kendall partial correlations with jackknifed standard errors


   %KENPAR(DATA=,X=,Y=,Z=,LIST=);

DATA (default _LAST_) is the dataset to be analyzed.

X and Y (REQUIRED) are the variables to be correlated.

Z (REQUIRED) is the variable to be partialed out.

LIST (default NO) indicaters whether you want a listing of the "leave-outs"
and their mean.


Sample program:

%INCLUDE "O:\BIOS162\SAMDAT.STP";
%INCLUDE "O:\BIOS256\KENPAR.MAC";
%KENPAR(DATA=SAMDAT,X=HEIGHT,Y=WEIGHT,Z=IQ,LIST=YES);


Output from sample program:

MACRO KENPAR: KENDALL PARTIAL CORRELATIONS

DATASET: SAMDAT  (25 RECORDS, 21 COMPLETE OBSERVATIONS)

X = HEIGHT,  Y = WEIGHT,  Z = IQ


                             TXY       TXZ       TYZ       TXY|Z     TXY - TXY|Z

CORRELATION COEFFICIENT      0.5513    0.0892    0.0907    0.5476    0.0037

X        Y        Z

53       57       105        0.5450    0.0952    0.1211    0.5399    0.0051
56       65       96         0.5211    0.1053    0.0895    0.5166    0.0045
57       98       118        0.5238    0.0635    0.0474    0.5224    0.0014

         (etc -- 15 lines of leave-outs omitted)

51       60       83         0.5263    0.0263    0.0579    0.5259    0.0005
50       54       85         0.5172    0.0158    0.0263    0.5170    0.0002
50       61       86         0.5368    0.0158    0.0474    0.5368    0.0001

         MEAN                0.5297    0.0859    0.0905    0.5256    0.0041

JACKKNIFED STANDARD ERROR    0.0938    0.1912    0.1890    0.0929    0.0147


| Note that KENPAR produces standard errors by jackknifing.  It also provides |
| a way to test the hypothesis of no difference between T(X,Y) and T(X,Y|Z).  |