Trailing-Edge
-
PDP-10 Archives
-
decus_20tap2_198111
-
decus/20-0026/probt.ssp
There are 2 other files named probt.ssp in the archive. Click here to see a list.
C PROB 10
C ..................................................................PROB 20
C PROB 30
C SUBROUTINE PROBT PROB 40
C PROB 50
C PURPOSE PROB 60
C TO OBTAIN MAXIMUM LIKELIHOOD ESTIMATES FOR THE PARAMETERS A PROB 70
C AND B IN THE PROBIT EQUATION Y = A + BX. AN ITERATIVE PROB 80
C SCHEME IS USED. THE INPUT TO THE SUBROUTINE CONSISTS OF K PROB 90
C DIFFERENT DOSAGE LEVELS APPLIED TO K GROUPS OF SUBJECTS, ANDPROB 100
C THE NUMBER OF SUBJECTS IN EACH GROUP RESPONDING TO THE PROB 110
C RESPECTIVE DOSAGE OF THE DRUG. PROB 120
C PROB 130
C USAGE PROB 140
C CALL PROBT (K,X,S,R,LOG,ANS,W1,W2,IER) PROB 150
C PROB 160
C DESCRIPTION OF PARAMETERS PROB 170
C K - NUMBER OF DIFFERENT DOSE LEVELS OF THE DRUG. K SHOULDPROB 180
C BE GREATER THAN 2. PROB 190
C X - INPUT VECTOR OF LENGTH K CONTAINING THE DOSE LEVEL OF PROB 200
C THE DRUG TESTED. X MUST BE NON-NEGATIVE. PROB 210
C S - INPUT VECTOR OF LENGTH K CONTAINING THE NUMBER OF PROB 220
C SUBJECTS TESTED AT EACH DOSE LEVEL PROB 230
C R - INPUT VECTOR OF LENGTH K CONTAINING THE NUMBER OF PROB 240
C SUBJECTS AT EACH LEVEL RESPONDING TO THE DRUG PROB 250
C LOG - INPUT OPTION CODE PROB 260
C 1- IF IT IS DESIRED TO CONVERT THE DOSE LEVELS TO PROB 270
C COMMON LOGARITHMS. THE DOSAGE LEVELS SHOULD BE PROB 280
C NON-NULL IN THIS CASE. PROB 290
C 0- IF NO CONVERSION IS DESIRED PROB 300
C ANS - OUTPUT VECTOR OF LENGTH 4 CONTAINING THE FOLLOWING PROB 310
C RESULTS PROB 320
C ANS(1)- ESTIMATE OF THE INTERCEPT CONSTANT A PROB 330
C ANS(2)- ESTIMATE OF THE PROBIT REGRESSION COEFFICIENT PROB 340
C B PROB 350
C ANS(3)- CHI-SQUARED VALUE FOR A TEST OF SIGNIFICANCE PROB 360
C OF THE FINAL PROBIT EQUATION PROB 370
C ANS(4)- DEGREES OF FREEDOM FOR THE CHI-SQUARE PROB 380
C STATISTIC PROB 390
C W1 - OUTPUT VECTOR OF LENGTH K CONTAINING THE PROPORTIONS PROB 400
C OF SUBJECTS RESPONDING TO THE VARIOUS DOSE LEVELS OF PROB 410
C THE DRUG PROB 420
C W2 - OUTPUT VECTOR OF LENGTH K CONTAINING THE VALUES OF THEPROB 430
C EXPECTED PROBIT FOR THE VARIOUS LEVELS OF A DRUG PROB 440
C IER - 1 IF K IS NOT GREATER THAN 2. PROB 450
C 2 IF SOME DOSAGE LEVEL IS NEGATIVE, OR IF THE INPUT PROB 460
C OPTION CODE LOG IS 1 AND SOME DOSAGE LEVEL IS ZERO. PROB 470
C 3 IF SOME ELEMENT OF S IS NOT POSITIVE. PROB 480
C 4 IF NUMBER OF SUBJECTS RESPONDING IS GREATER THAN PROB 490
C NUMBER OF SUBJECTS TESTED. PROB 500
C ONLY IF IER IS ZERO IS A PROBIT ANALYSIS PERFORMED. PROB 510
C OTHERWISE, ANS, W1, AND W2 ARE SET TO ZERO. PROB 520
C PROB 530
C REMARKS PROB 540
C THE PROGRAM WILL ITERATE ON THE PROBIT EQUATION UNTIL TWO PROB 550
C SUCCESSIVE SOLUTIONS PRODUCE CHANGES OF LESS THAN 10**(-7). PROB 560
C PROB 570
C SUBROUTINES AND FUNCTION SUBPROGRAMS REQUIRED PROB 580
C NDTR PROB 590
C NDTRI PROB 600
C PROB 610
C METHOD PROB 620
C REFER TO D. J. FINNEY, PROBIT ANALYSIS. 2ND ED. (CAMBRIDGE, PROB 630
C 1952) PROB 640
C PROB 650
C ..................................................................PROB 660
C PROB 670
SUBROUTINE PROBT (K,X,S,R,LOG,ANS,W1,W2,IER) PROB 680
C PROB 690
DIMENSION X(1),S(1),R(1),ANS(1),W1(1),W2(1) PROB 700
C PROB 710
C TEST WHETHER LOG CONVERSION IS NEEDED PROB 720
C PROB 730
IER=0 PROB 740
IF(K-2)5,5,7 PROB 750
5 IER = 1 PROB 760
GO TO 90 PROB 770
7 DO 8 I=1,K PROB 780
IF(X(I))12,8,8 PROB 790
8 CONTINUE PROB 800
IF(LOG-1) 16,10,16 PROB 810
10 DO 15 I=1,K PROB 820
IF(X(I))12,12,14 PROB 830
12 IER=2 PROB 840
GO TO 90 PROB 850
14 X(I)= ALOG10(X(I)) PROB 860
15 CONTINUE PROB 870
C PROB 880
C COMPUTE PROPORTIONS OF OBJECTS RESPONDING PROB 890
C PROB 900
16 DO 18 I=1,K PROB 910
IF(S(I)-R(I)) 17,18,18 PROB 920
17 IER=4 PROB 930
GO TO 90 PROB 940
18 CONTINUE PROB 950
20 DO 23 I=1,K PROB 960
IF(S(I))21,21,22 PROB 970
21 IER=3 PROB 980
GO TO 90 PROB 990
22 W1(I)=R(I)/S(I) PROB1000
23 CONTINUE PROB1010
C PROB1020
C COMPUTE INITIAL ESTIMATES OF INTERCEPT AND PROBIT REGRESSION PROB1030
C COEFFICIENT PROB1040
C PROB1050
WN=0.0 PROB1060
XBAR=0.0 PROB1070
SNWY=0.0 PROB1080
SXX=0.0 PROB1090
SXY=0.0 PROB1100
C PROB1110
DO 30 I=1,K PROB1120
P=W1(I) PROB1130
IF(P) 30, 30, 24 PROB1140
24 IF(P-1.0) 25, 30, 30 PROB1150
25 WN=WN+1.0 PROB1160
C PROB1170
CALL NDTRI (P,Z,D,IER) PROB1180
C PROB1190
Z=Z+5.0 PROB1200
XBAR=XBAR+X(I) PROB1210
SNWY=SNWY+Z PROB1220
SXX=SXX+X(I)**2 PROB1230
SXY=SXY+X(I)*Z PROB1240
30 CONTINUE PROB1250
C PROB1260
B=(SXY-(XBAR*SNWY)/WN)/(SXX-(XBAR*XBAR)/WN) PROB1270
XBAR=XBAR/WN PROB1280
SNWY=SNWY/WN PROB1290
A=SNWY-B*XBAR PROB1300
DD=0.0 PROB1310
C PROB1320
C COMPUTE EXPECTED PROBIT PROB1330
C PROB1340
DO 31 I=1,K PROB1350
31 W2(I)=A+B*X(I) PROB1360
C PROB1370
33 SNW=0.0 PROB1380
SNWX=0.0 PROB1390
SNWY=0.0 PROB1400
SNWXX=0.0 PROB1410
SNWXY=0.0 PROB1420
DO 50 I=1,K PROB1430
Y=W2(I) PROB1440
C PROB1450
C FIND A WEIGHTING COEFFICIENT FOR PROBIT ANALYSIS PROB1460
C PROB1470
D=Y-5.0 PROB1480
C PROB1490
CALL NDTR (D,P,Z) PROB1500
C PROB1510
Q=1.0-P PROB1520
W=(Z*Z)/(P*Q) PROB1530
C PROB1540
C COMPUTE WORKING PROBIT PROB1550
C PROB1560
IF(Y-5.0) 35, 35, 40 PROB1570
35 WP=(Y-P/Z)+W1(I)/Z PROB1580
GO TO 45 PROB1590
40 WP=(Y+Q/Z)-(1.0-W1(I))/Z PROB1600
C PROB1610
C SUM INTERMEDIATE RESULTS PROB1620
C PROB1630
45 WN=W*S(I) PROB1640
SNW=SNW+WN PROB1650
SNWX=SNWX+WN*X(I) PROB1660
SNWY=SNWY+WN*WP PROB1670
SNWXX=SNWXX+WN*X(I)**2 PROB1680
50 SNWXY=SNWXY+WN*X(I)*WP PROB1690
C PROB1700
C COMPUTE NEW ESTIMATES OF INTERCEPT AND COEFFICIENT PROB1710
C PROB1720
XBAR=SNWX/SNW PROB1730
C PROB1740
SXX=SNWXX-(SNWX)*(SNWX)/SNW PROB1750
SXY=SNWXY-(SNWX)*(SNWY)/SNW PROB1760
B=SXY/SXX PROB1770
C PROB1780
A=SNWY/SNW-B*XBAR PROB1790
C PROB1800
C EXAMINE THE CHANGES IN Y PROB1810
C PROB1820
SXX=0.0 PROB1830
DO 60 I=1,K PROB1840
Y=A+B*X(I) PROB1850
D=W2(I)-Y PROB1860
SXX=SXX+D*D PROB1870
60 W2(I)=Y PROB1880
IF(( ABS(DD-SXX))-(1.0E-7)) 65, 65, 63 PROB1890
63 DD=SXX PROB1900
GO TO 33 PROB1910
C PROB1920
C STORE INTERCEPT AND COEFFICIENT PROB1930
C PROB1940
65 ANS(1)=A PROB1950
ANS(2)=B PROB1960
C PROB1970
C COMPUTE CHI-SQUARE PROB1980
C PROB1990
ANS(3)=0.0 PROB2000
DO 70 I=1,K PROB2010
Y=W2(I)-5.0 PROB2020
C PROB2030
CALL NDTR (Y,P,D) PROB2040
C PROB2050
AA=R(I)-S(I)*P PROB2060
DD=S(I)*P*(1.0-P) PROB2070
70 ANS(3)=ANS(3)+AA*AA/DD PROB2080
C PROB2090
C DEGREES OF FREEDOM FOR CHI-SQUARE PROB2100
C PROB2110
ANS(4)=K-2 PROB2120
C PROB2130
80 RETURN PROB2140
90 DO 100 I=1,K PROB2150
W1(I)=0.0 PROB2160
100 W2(I)=0.0 PROB2170
DO 110 I=1,4 PROB2180
110 ANS(I)=0.0 PROB2190
GO TO 80 PROB2200
END PROB2210