EXST7015 : Statistical Techniques II Geaghan Contrasts Experimental Design examples Page 1 19g-Anova_ContrastExamples.doc 1 **EXAMPLE 4*****************************************************; 2 *** Example of RBD ***; 3 *** From Snedecor & Cochran, 1980 (pg 256) ***; 4 ****************************************************************; 5 OPTIONS PS=256 LS=78 NOCENTER NODATE PAGENO=1; 6 DATA SOYBEAN; INFILE CARDS MISSOVER; 7 INPUT TREATMNT $ BLOCK FAILURES; 8 TITLE1 'FAILURES TO GERMINATE OF SOYBEAN PLANTS'; 9 TITLE2 '4 TREATMENTS AND A CONTROL, 5 BLOCKS'; 10 CARDS; 36 ; 37 PROC MIXED DATA=SOYBEAN cl COVTEST; CLASSES TREATMNT BLOCK; 38 TITLE3 'ANOVA with PROC MIXED - RBD without reps'; 39 MODEL FAILURES = TREATMNT / htype=3 DDFM=Satterthwaite outp=ResidDataP; 40 RANDOM BLOCK; 41 *** Treatment levels ------------ ARASAN CHECK FERMATE SEMESAN SPERGON; 42 CONTRAST 'Check v othrs' TREATMNT 1 -4 1 1 1! ; 43 CONTRAST 'S v others' TREATMNT -1 0 -1 1 1; 44 CONTRAST 'A v others' TREATMNT -3 0 1 1 1; 45 CONTRAST 'F v others' TREATMNT 1 0 -3 1 1; 46 RUN; NOTE: Convergence criteria met. NOTE: The data set WORK.RESIDDATAP has 25 observations and 10 variables. NOTE: The PROCEDURE MIXED printed page 1. NOTE: PROCEDURE MIXED used: real time 0.23 seconds cpu time 0.18 seconds 47 FAILURES TO GERMINATE OF SOYBEAN PLANTS 1 4 TREATMENTS AND A CONTROL, 5 BLOCKS ANOVA with PROC MIXED - RBD without reps Class Level Information Class Levels Values TREATMNT 5 ARASAN CHECK FERMATE SEMESAN SPERGON BLOCK 5 1 2 3 4 5 Covariance Parameter Estimates Standard Z Cov Parm Estimate Error Value Pr Z Alpha Lower Upper BLOCK 1.4100 1.8032 0.78 0.2171 0.05 0.3075 430.52 Residual 5.4100 1.9127 2.83 0.0023 0.05 3.0008 12.5310 Fit Statistics -2 Res Log Likelihood 101.9 AIC (smaller is better) 105.9 AICC (smaller is better) 106.6 BIC (smaller is better) 105.1 Type 3 Tests of Fixed Effects Effect Num DF Den DF F Value Pr > F TREATMNT 4 16 3.87 0.0219 Contrasts Label Num DF Den DF F Value Pr > F Check v othrs 1 16 12.43 0.0028 S v others 1 16 1.81 0.1971 A v others 1 16 0.31 0.5865 F v others 1 16 1.00 0.3326EXST7015 : Statistical Techniques II Geaghan Contrasts Experimental Design examples Page 2 19g-Anova_ContrastExamples.doc 48 **EXAMPLE 5********************************************; 49 *** Example of RBD with sampling error ***; 50 *** From Snedecor & Cochran, 1980 (pg 267) ***; 51 *******************************************************; 52 OPTIONS PS=256 LS=78 NOCENTER NODATE PAGENO=1; 53 DATA FUMIGANT; INFILE CARDS MISSOVER; 54 INPUT FUMIGANT $ BLOCK $ W1 W2 W3 W4; 55 TITLE1 'Number of wire worms found for 2 fumigants and a control'; 56 TITLE2 'Fumigants are C and S, control is 0, 5 BLOCKS'; 57 REP=1; WORMS=W1; LWORMS=LOG(WORMS+1); OUTPUT; 58 REP=2; WORMS=W2; LWORMS=LOG(WORMS+1); OUTPUT; 59 REP=3; WORMS=W3; LWORMS=LOG(WORMS+1); OUTPUT; 60 REP=4; WORMS=W4; LWORMS=LOG(WORMS+1); OUTPUT; 61 KEEP FUMIGANT BLOCK REP WORMS LWORMS; 62 CARDS; 79 PROC mixed DATA=FUMIGANT cl; CLASSES FUMIGANT BLOCK REP; 80 TITLE3 'ANOVA with PROC MIXED - RBD with reps'; 81 MODEL WORMS = FUMIGANT / htype=3 DDFM=Satterthwaite outp=ResidDataP outpM=ResidDataPM; 82 RANDOM BLOCK FUMIGANT*BLOCK; 83 *** FUMIGANT levels ---------------- 0 C S; 84 CONTRAST 'Control v othrs' FUMIGANT -2 1 1; 85 CONTRAST 'C v S' FUMIGANT 0 -1 1; 86 RUN; Number of wire worms found for 2 fumigants and a control 1 Fumigants are C and S, control is 0, 5 BLOCKS ANOVA with PROC MIXED - RBD with reps Class Level Information Class Levels Values FUMIGANT 3 0 C S BLOCK 5 I II III IV V REP 4 1 2 3 4 Covariance Parameter Estimates Cov Parm Estimate Alpha Lower Upper BLOCK 1.1052 0.05 0.1473 25730931 FUMIGANT*BLOCK 3.8559 0.05 1.2517 50.5437 Residual 9.1056 0.05 6.2643 14.4450 Fit Statistics -2 Res Log Likelihood 310.3 AIC (smaller is better) 316.3 AICC (smaller is better) 316.7 BIC (smaller is better) 315.1 Type 3 Tests of Fixed Effects Effect Num DF Den DF F Value Pr > F FUMIGANT 2 8 5.98 0.0258 Contrasts Label Num DF Den DF F Value Pr > F Control v othrs 1 8 11.88 0.0087 C v S 1 8 0.08 0.7812EXST7015 : Statistical Techniques II Geaghan Contrasts Experimental Design examples Page 3 19g-Anova_ContrastExamples.doc 89 **EXAMPLE 6********************************************; 90 *** Example of Latin Square Design ***; 91 *** From Snedecor & Cochran, 1980 (pg 271) ***; 92 *******************************************************; 93 OPTIONS PS=256 LS=78 NOCENTER NODATE PAGENO=1; 94 DATA MILLET; INFILE CARDS MISSOVER; 95 INPUT ROW COLUMN Spacing $ YIELD; 96 TITLE1 'LATIN SQUARE WITH 5 ROWS, COLUMNS AND TREATMENTS'; 97 TITLE2 'MILLET YIELDS (G) FOR SPACINGS OF 2, 4, 6, 8 AND 10 INCHES'; 98 CARDS; NOTE: The data set WORK.MILLET has 25 observations and 4 variables. NOTE: DATA statement used: real time 0.03 seconds cpu time 0.03 seconds 124 ; 125 PROC MIXED