untitled

the replication. Thus its influence is factorized out in the first option or randomized

across treatments in the second.

8 All of these rules may be broken, but that degrades the design to one yielding

neither strong inference nor an unambiguous conclusion. Such results are still use-

ful so long as their dubious nature is fully appreciated and declared. Environmental

impact assessments are just such examples.

288 Chapter 16

June 1987 June 1988

Day 1 Day 2 Day 3 Day 1 Day 2 Day 3

Red kangaroos 45 19 18 72 31 28

17 51 44 32 29 9

8 8 61 94 34 48

28 11 35 27 47 38

26 72 65 66 21 91

48 34 76 55 49 138

53 67 52 102 29 67

___^62 ___^27 ___^30 ___^41 ___^67106 ___

Tijk= 287 289 381 489 307 525

x ̄= 35.9 36.1 47.6 61.1 38.4 65.6

Gray kangaroos 66 27 27 116 65 10

52 47 66 81 57 22

34 13 75 63 41 43

8 16 104 25 33 63

35 101 109 75 120 74

36 150 170 77 28 101

42 116 51 59 62 76

___^65 ___^66 ___^14 ___^59 ___^17 ___^82

Tijk= 338 536 616 555 423 471

x ̄= 42.3 67.0 77.0 69.4 52.9 58.9

ANOVA

Source SS d.f. MS FF0.05*

ROW(SPECIES) 4,551.3 1 4551.3 4.7 4.0

COLUMN(DAYS) 3,227.3 2 1613.6 1.6 3.1

LAYER(YEARS) 1,086.8 1 1086.8 1.1 4.0

RC interaction 1,018.1 2 509.0 0.5 3.1

CL interaction 4,681.6 2 2340.8 2.3 3.1

RL interaction 1,708.6 1 1708.6 1.7 4.0

RCL interaction 1,444.7 2 722.3 0.7 3.1

Residual 86,359.4 84 1028.1

Total 104,077.7 95

*F0.05is the critical level that must be exceeded by the observed Fto qualify for a probability equal

to or less than 5%.

Box 16.3Red

kangaroos and gray

kangaroos counted on

the Cunnamulla degree

block (10,870 km^2 ) in

1987 and 1988. Each

replicate is the number

of kangaroos counted

on a transect measuring

0.4 km by 90 km.

WECC16 18/08/2005 14:47 Page 288