C CHIS 10
C ..................................................................CHIS 20
C CHIS 30
C SUBROUTINE CHISQ CHIS 40
C CHIS 50
C PURPOSE CHIS 60
C COMPUTE CHI-SQUARE FROM A CONTINGENCY TABLE CHIS 70
C CHIS 80
C USAGE CHIS 90
C CALL CHISQ(A,N,M,CS,NDF,IERR,TR,TC) CHIS 100
C CHIS 110
C DESCRIPTION OF PARAMETERS CHIS 120
C A - INPUT MATRIX, N BY M, CONTAINING CONTINGENCY TABLE CHIS 130
C N - NUMBER OF ROWS IN A CHIS 140
C M - NUMBER OF COLUMNS IN A CHIS 150
C CS - CHI-SQUARE (OUTPUT) CHIS 160
C NDF - NUMBER OF DEGREES OF FREEDOM (OUTPUT) CHIS 170
C IERR - ERROR CODE (OUTPUT) CHIS 180
C 0 - NORMAL CASE CHIS 190
C 1 - EXPECTED VALUE IS LESS THAN 1.0 IN ONE OR CHIS 200
C MORE CELLS CHIS 210
C 3 - NUMBER OF DEGREES OF FREEDOM IS ZERO CHIS 220
C TR - WORK VECTOR OF LENGTH N CHIS 230
C TC - WORK VECTOR OF LENGTH M CHIS 240
C CHIS 250
C REMARKS CHIS 260
C IF ONE OR MORE CELLS CONTAIN AN EXPECTED VALUE (I.E., CHIS 270
C THEORETICAL VALUE) LESS THAN 1.0, CHI-SQUARE WILL BE CHIS 280
C COMPUTED, BUT ERROR CODE WILL BE SET TO 1. CHIS 290
C SEE REFERENCE GIVEN BELOW. CHIS 300
C CHI-SQUARE IS SET TO ZERO IF EITHER N OR M IS ONE (ERROR CHIS 310
C CODE 3). CHIS 320
C CHIS 330
C SUBROUTINES AND FUNCTION SUBPROGRAMS REQUIRED CHIS 340
C NONE CHIS 350
C CHIS 360
C METHOD CHIS 370
C DESCRIBED IN S. SIEGEL, 'NONPARAMETRIC STATISTICS FOR THE CHIS 380
C BEHAVIORAL SCIENCES', MCGRAW-HILL, NEW YORK, 1956, CHIS 390
C CHAPTER 6 AND CHAPTER 8. CHIS 400
C CHIS 410
C ..................................................................CHIS 420
C CHIS 430
SUBROUTINE CHISQ(A,N,M,CS,NDF,IERR,TR,TC) CHIS 440
DIMENSION A(1),TR(1),TC(1) CHIS 450
C CHIS 460
NM=N*M CHIS 470
IERR=0 CHIS 480
CS=0.0 CHIS 490
C CHIS 500
C FIND DEGREES OF FREEDOM CHIS 510
C CHIS 520
NDF=(N-1)*(M-1) CHIS 530
IF(NDF) 5,5,10 CHIS 540
5 IERR=3 CHIS 550
RETURN CHIS 560
C CHIS 570
C COMPUTE TOTALS OF ROWS CHIS 580
C CHIS 590
10 DO 90 I=1,N CHIS 600
TR(I)=0.0 CHIS 610
IJ=I-N CHIS 620
DO 90 J=1,M CHIS 630
IJ=IJ+N CHIS 640
90 TR(I)=TR(I)+A(IJ) CHIS 650
C CHIS 660
C COMPUTE TOTALS OF COLUMNS CHIS 670
C CHIS 680
IJ=0 CHIS 690
DO 100 J=1,M CHIS 700
TC(J)=0.0 CHIS 710
DO 100 I=1,N CHIS 720
IJ=IJ+1 CHIS 730
100 TC(J)=TC(J)+A(IJ) CHIS 740
C CHIS 750
C COMPUTE GRAND TOTAL CHIS 760
C CHIS 770
GT=0.0 CHIS 780
DO 110 I=1,N CHIS 790
110 GT=GT+TR(I) CHIS 800
C CHIS 810
C COMPUTE CHI SQUARE FOR 2 BY 2 TABLE (SPECIAL CASE) CHIS 820
C CHIS 830
IF(NM-4) 130,120,130 CHIS 840
120 CS=GT*(ABS(A(1)*A(4)-A(2)*A(3))-GT/2.0)**2 /(TC(1)*TC(2)*TR(1) CHIS 850
1*TR(2)) CHIS 860
RETURN CHIS 870
C CHIS 880
C COMPUTE CHI SQUARE FOR OTHER CONTINGENCY TABLES CHIS 890
C CHIS 900
130 IJ=0 CHIS 910
DO 140 J=1,M CHIS 920
DO 140 I=1,N CHIS 930
IJ=IJ+1 CHIS 940
E=TR(I)*TC(J)/GT CHIS 950
IF(E-1.0) 135, 140, 140 CHIS 960
135 IERR=1 CHIS 970
140 CS=CS+(A(IJ)-E)*(A(IJ)-E)/E CHIS 980
RETURN CHIS 990
END CHIS1000