Google
 

Trailing-Edge - PDP-10 Archives - decuslib10-02 - 43,50145/krank.doc
There are 2 other files named krank.doc in the archive. Click here to see a list.
SUBROUTINE KRANK

PURPOSE
   TEST CORRELATION BETWEEN TWO VARIABLES BY MEANS OF KENDALL
   RANK CORRELATION COEFFICIENT

USAGE
   CALL KRANK(A,B,R,N,TAU,SD,Z,NR)

DESCRIPTION OF PARAMETERS
   A   - INPUT VECTOR OF N OBSERVATIONS FOR FIRST VARIABLE
   B   - INPUT VECTOR OF N OBSERVATIONS FOR SECOND VARIABLE
   R   - OUTPUT VECTOR OF RANKED DATA OF LENGTH 2*N. SMALLEST
	 OBSERVATION IS RANKED 1, LARGEST IS RANKED N. TIES
	 ARE ASSIGNED AVERAGE OF TIED RANKS.
   N   - NUMBER OF OBSERVATIONS
   TAU - KENDALL RANK CORRELATION COEFFICIENT (OUTPUT)
   SD  - STANDARD DEVIATION (OUTPUT)
   Z   - TEST OF SIGNIFICANCE OF TAU IN TERMS OF NORMAL
	 DISTRIBUTION (OUTPUT)
   NR  - CODE, 0 FOR UNRANKED DATA IN A AND B, 1 FOR RANKED
	 DATA IN A AND B (INPUT)

REMARKS
   SD AND Z ARE SET TO ZERO IF N IS LESS THAN TEN

SUBROUTINES AND FUNCTION SUBPROGRAMS REQUIRED
   RANK
   TIE

METHOD
   DESCRIBED IN S. SIEGEL, 'NONPARAMETRIC STATISTICS FOR THE
   BEHAVIORAL SCIENCES', MCGRAW-HILL, NEW YORK, 1956,
   CHAPTER 9