Agriculture Reference
In-Depth Information
This is a dataset of 10 tomato plants where fungal strains A and
B are inoculated on two different leaves on each plant (Clewer and
Scarisbrick, 2001, p. 298). The number of fungal colonies that develop
is then counted. In this case, with a probability > |z| equal to 0.1132,
the null hypothesis that there are no differences between the fungal
strains cannot be ruled out.
The Mann-Whitney Test is a nonparametric procedure for un-
matched data. In this case, it is the ranking of the data that is important
to the analysis rather than the actual data. Although this is a nonpara-
metric approach, there are some underlying assumptions that the popu-
lations are similar, but not necessarily normal with similar variances.
This test is the nonparametric equivalent of the independent sample
t-test. An experiment was conducted comparing a standard wheat
variety to a new variety in a CRD (completely randomized design) with
10 plots for the standard variety and 6 plots for the new variety (Clewer
and Scarisbrick, 2001, p. 300). Open the dataset Wheat Variety Test.dta
and enter the following command to see the results:
ranksum
yield,
by(
variety
)
Two-sample Wilcoxon rank-sum (Mann-Whitney) test
variety | obs rank sum expected
-------------+---------------------------------
1 | 6 71 51
2 | 10 65 85
-------------+---------------------------------
combined | 16 136 136
unadjusted variance 85.00
adjustment for ties -0.88
----------
adjusted variance 84.12
Ho: yield(variety==1) = yield(variety==2)
z = -2.181
Prob > |z| = 0.0292
The results indicate that the null hypothesis of equal yield should be
rejected with a Prob > |z| = 0.0292. The median of the new wheat vari-
ety is 2.35 t/ha, while the standard variety had a median of 2.0 t/ha.
Search WWH ::
Custom Search