Post Hoc Tests - Agricultural Statistical Data Analysis Using Stata - page 149

test 2.trt - 3.trt = 0
( 1) 2.trt - 3.trt = 0
F( 1, 9) = 1.49
Prob > F = 0.2525
test 2.trt + 3.trt - 2*4.trt = 0
( 1) 2.trt + 3.trt - 2*4.trt = 0
F( 1, 9) = 8.45
Prob > F = 0.0174
The number preceding the variable reflects the level of the vari-
able. This dataset has value labels associated with the levels of trt
indicating what the treatments were. To see the actual levels (i.e.,
numbers) associated with trt , enter label list in the Command
window. The above comparisons can be entered with different alge-
braic expressions and have the same result; however, this should be
avoided to prevent mistakes in the contrasts. Entering them as they
appear in Table 8.1 as orthogonal contrasts is a good habit to get into.
With a slightly more complex situation, load the dataset Corn Seed
Treatments.dta, which is a dataset of corn stand counts from a green-
house experiment of different fungicide seed treatments (Steele and
Torrie, 1980, p. 206). * This dataset has eight treatments that include
an untreated check, two types of mercuric fungicides, two types of
nonmercuric fungicides from one company and three types of non-
mercuric fungicides from a second company. The last three treatments
include different formulations of the same material. The planned
comparisons are shown in Table 8.2.
After loading the dataset, run an ANOVA ( anova stand trt
rep ). The post hoc tests and the results of these planned comparisons
are
test 7*1.trt - 2.trt - 3.trt - 4.trt - 5.trt - 6.trt - 7.trt
- 8.trt = 0
( 1) 7*1b.trt - 2.trt - 3.trt - 4.trt - 5.trt - 6.trt - 7.trt
- 8.trt = 0
F( 1, 35) = 4.95
Prob > F = 0.0327
* Mercury-based seed treatments are no longer allowed in agricultural production.