TO STEVO VN TET SNE SM AE, ME aja

The CONTRAST subcommand indicates the contrast coefficients for factor CEFFECT. The

PARTITION subcommand divides the CEFFECT factor into 2 groups for the contrasts. The first

group has one degree of freedom with the contrast (3, -1, -1, -1). The second group (CEFFECT(2))

corresponds to the second and third contrasts lumped together and has two degrees of freedom.
The complete MANOVA command file is given in Figure 1.28a.

Figure 1.28a
RUN NAME ANALYSIS OF VARIANCE WITH CARRY-OVER EFFECTS.
COMMENT DATA IS TAKEN FROM COCHRAN AND COX(1957) PAGE 135.

CEFFECT REPRESENTS THE CARRY-OVER EFFECTS.
CEFFECT<l IF NO RESIDUAL EFFECTS.
2 IF RESIDUAL EFFECT A.
3 IF RESIDVAL EFFECT B.
4 IF RESIDUAL EFFECT C.
VARIABLE LIST  PERIOD,COW, SOUARE, TREATMNT, CEFFECT , DEP
INPUT MEDIUM CARD
INPUT FORMAT  FIXED(2X,5Fi.0,F1l0.0)
N OF CASES 18
MANOVA DEP BY PERIOD(1,3),C0W(1,6),SOUARE(1,2),
TREATMNT(1,3). CEFFECT(1,4)/

CONTRAST(CEFFECT) < SPECIAL(l l l i, 3 -l -il -l,
O 2-1-I, 00 1-i )/

PARTITION(CEFFECT) < (1,2)/

DESIGN < COW, PERIOD WITHIN SOUARE,
CEFFECT(2), TREATMNT/

DESIGN <- COW, PERIOD WITHIN SOVARE,
TREATMNT, CEFFECT(2)/

i

READ INPUT DATA
l

lili 38.
1212. 109.
13131 124.
i421ll 86.
15221 75.
16231 101.
21122 25.
22133 86.
23114 TR.
24232 T6.
25213 35.
26224 63.
31133 15.
S2ll4 39.
33122 27.
34224 46.
35232 54.
36213 l.
FINISH

no
In the first DESIGN specification, treatment effects are adjusted for the residual effects, and

the converse holds in the second DESIGN specification. The ANOVA summary tables for both
models are given in Figure 1.28b.

Figure 1.28b
a o NE ENNNN
TESTS OF SIGNIFICANCE FOR DEP USING SEGUENTIAL SUMS OF SOUARES

SOURCE OF VARIATION SUM OF SOUARES DF MEAN SOUARE F SIG. OF F
RESIDUAL 199,25000 4 49.81250

CONSTANT 61483.55556 li  861483.55556 1234.29974 0.0
CoOW 5781.1111l 5 1156.22222 23.21149 005
PERIOD WITHIN SOUARE 11489.11111l 4 2872.27778 57.66179 -001
CEFFECT(2) 38.42222 2 i9.211ll 38567 103
TREATMNT 2854 .55000 2 1427.27500 . 28.65295 004

TESTS OF SIGNIFICANCE FOR DEP USING SEGUENTIAL SUMS OF SOVARES

SOURCE OF VARIATION SUM OF SOUARES DF MEAN SOUARE F SIG. OF F

RESIDUAL 199.25000 4 49.81250

CONSTANT 61483 .55556 l  61483.55556 1254.29974 0.0
OW ST81.11111 5 1156.22222 23.21149 005
PERIOD WITHIN SOUARE 11489.11111 4 2872..27778 57.66179 -001
TREATMNT 2276.TTTT8 2 1138.38889 22..85348 006
CEFFECT(2) 616. 19444 2 308.09722 6. 18514 -060

Note that in this example, the number of observations receiving r,, r, and r, are egual (4). If
the design is not balanced with respect to residual effects, contrast coefficients for unegual numbers
of replicates must be used to create the desired residual effects.

MANOVA 29