Agriculture Reference
In-Depth Information
kwallis yield, by ( trt )
Kruskal-Wallis equality-of-populations rank test
+---------------------------------+
| trt | Obs | Rank Sum |
|----------------+-----+----------|
| Dol-Mix (1 kg) | 4 | 65.50 |
| Dol-Mix (2 kg) | 4 | 97.00 |
| DDT + y-BHC | 4 | 92.00 |
| Azodrin | 4 | 63.50 |
| Dimecron-Boom | 4 | 41.00 |
|----------------+-----+----------|
| Dimecron-Knap | 4 | 35.00 |
| Control | 4 | 12.00 |
+---------------------------------+
chi-squared = 21.050 with 6 d.f.
probability = 0.0018
chi-squared with ties = 21.061 with 6 d.f.
probability = 0.0018
Look at these results compared to the previous Kruskal-Wallis
analysis. The chi-square with ties in the second analysis is slightly dif-
ferent than the chi-square without ties (21.050 vs. 21.061). In the first
analysis, these values are the same (7.269). Because we are dealing
with ranked data, ties will have an impact on the outcome.
Friedman's Test is a nonparametric test for two-way classified
data, such as would be found in an RCBD (randomized complete
block design) where the treatments are one factor and blocking or
replications are another. This test is not directly available in Stata,
but can be downloaded as an ado file. Use the findit command
to locate the friedman command (remember, you have to have
an Internet connection). Install this program and open the dataset
Flaxseed Oil Content 2.dta. This dataset is of an experiment evaluat-
ing Redwing flaxseed oil content based on growth stage of inocula-
tion with Septoria linocola, the causal organism of pasmo (Steel and
Torrie, 1980, p. 547).
he friedman command, unlike most commands in Stata, requires
that each block or replication occur as a separate variable. This is often
the way such data are presented in textbooks, rather than the conven-
tion of blocking or replication occurring as a single variable label. After
opening the dataset, enter the command and see the results.
Search WWH ::




Custom Search