APPENDIX I 

 PPSORT Program 



SJBROUTI NE PPS JKT ( NNM, I D .CHAR, NNS AM, DS AM, PP , T X , TS , T P , N SAM, N , L R , 

 1 MK.KR) 



ni 'lE.NSICN VFC (500) ,CP( 500) , i:jlJ (500 ),CCP (500) 

 C THE ABOVE FCUR VECTORS MUST BE DMENSIONED GREATER THAN NNM + 2 



DI MENS I ON ID ( mU) , CHAR (NNN) , I OS AM (NINS A^') , PP (KNS AM ) , NR ( 1 2B I 



EOUIVALENCE ( VECCCP) 

 C MLR,MMR,MKR, AND NR ARE USEi) IN THE RANDOM NUMBER GENERATOR. 



C TO START THE RANDOM NUMBER GENERATOR, THREE ODD RANDOM INTEGERS 



C SHOULD 8E ASSIGNED TC LR , KR , AND MR, RESPECTIVELY. 



C SEE THE REFERENCE CITED BELOW. 



DATA MLR, Mf'R, MKR / 65539, 33554^^33, 362436069/ 



DATA NR / 



1 



?2I , 



301 , 



10 ^, 



2H1 , 



445 , 



273 





475, 



437, 



277, 



353, 



153, 



2 37, 



465, 



2 



61 , 



2>37, 



413 , 



44 1 , 



369, 



99 





51, 



463, 



179, 



55, 



4S3, 



95, 



363, 



3 



437 , 



253, 



197, 



409, 



303, 



231 





435, 



209, 



357, 



143, 



205, 



255, 



91, 





3, 



131 , 



409, 



399, 



371, 



453 





97, 



271, 



307, 



31, 



123, 



243, 



29, 



5 



343 , 



217, 



391 , 



35, 



337, 



229 





49 3 , 



263, 



317, 



75, 



449, 



45, 



279, 



6 



245 , 



32 3 , 



57, 



191 , 



17, 



133 





73, 



339, 



18 7, 



165, 



137, 



121, 



365, 



7 



33 , 



289, 



473 , 



405 , 



3 01 , 



77 





48 1 , 



351, 



159, 



349, 



295, 



49, 



359, 



8 



393, 



19, 



265, 



2cJ5, 



7, 



167 





451, 



479, 



10 3, 



257, 



459, 



93, 



331, 



9 



149, 



203, 



325 , 



215, 



259, 



33 





115, 



233, 



145, 



117, 



139, 



9, 



431, 



X 



319, 



141 , 



443 , 



129, 



13, 



71 





38 3 , 



85, 



305, 



63, 



379 



/ 





N = NNN 

 NSAM = NNSAV 



C THIS PROGRAM WILL DRAW 'NSAM* SAMPLES FROM THE' ID* POPULATION OF 



C 'N' UNITS WITH PROBABILITY PROPCRTICNAL TC TFE ASSOCIATED VALUE 



C OF 'CHAR' WITHOUT REPLACEMENT. 



C THE SAMPLE IS DRAWN S YS TE f< AT I C AL L Y FROM THE LIST ORDERED BY 



C INCREASING VALJES OF 'CHAR', USING THE THEORY CESCRIBED IN: 



C HARTLEY, H.O.; JCUR. A^ER. STAT. ASSOC. 6 I ( 3 15 ): 739- 748. 



C 



C THE UNITS SELECTED WILL BE IDENTIFIED IN THE VECTOR 'IDSAM'. 



C 



C 'JNSAM' MAY BE DECREASED SU THAT THE STEP IS GREATER THAN THE 



C MAXIMUM VALUE OF 'CHAR'. IN THAT EVENT, 



C THE LARGEST ELEMENT OF THE POPULATION WILL BE ASSIGNED A SAMPLING 



C PROBABILITY OF UNITY AND DELETED FROM THE POPULATION. 



C 



C •□' IS THE STEP SIZE USED I.J THE SYSTEMATIC SELECTION. 



C 



C FOR PURPOSES OF ERROR CALCULATICN, THE SUM OF SQUARED DEVIATIONS 



C OF THE MEAN OF CHAR FOR EACH OF THE 'N' POSSIBLE SAMPLES 



C FROM THE WEIGHTED MEAN OF THE SAMPLE MEANS, E/SCH WEIGHTED BY THE 



C PROBABILITY WITH WHICH THAT SAMPLE CLUSTER MAY BE SELECTED, IS 



C COMPUTED IN 'TS'. ONE MINUS THE SUM GF SQUARES OF THE PROBABILITY 



C IS CALCULATED IN 'TP'. 



C THE REMAINING "ARTS OF THE EXPRESSICN FOR SAMPLING ERROR MUST BE 



C CALCULATED AFTER THE SAMPLE IS MEASURED. (SEE HARTLEY'S EQ.(29)). 



C 



IF ( N .LT. NSAM) NSAM = N 

 IF ( NSAM .LE. ) RETURN 

 VEC( 1 ) =0.0 

 IND(N+1) = 1 

 CHAR( N+ 1) = 0.0 



C FIRST SORT INT) INCREASING ORDER CF CHAR WITHCUT RE-ARRANGING 



DO 2 J = 1 , N 

 2 INO(J) = J 

 NMl = N - 1 



13 



