Agriculture Reference
In-Depth Information
In this case, the difference between the two varieties, PMK-06-04
and Orange Bulldog, is not significant. The output lists three different
probabilities, T<t, |T|>|t|, and T>t. The first and last values, 0.3447
and 0.6553, are one-tailed probability values for the t-test, while the
middle value, 0.6895, is a two-tailed probability value. Note that the
first and third probability values add to 1.
There are two other forms of the t-test available: the paired and
unpaired. To see these forms, open the file ttestpumpkin.dta and look
at the data. It is the same data as in simplepumpkin.dta, but the data
have been entered in two columns. If the command is entered
ttest
normal
==
new
, it is assumed to be paired and, although in this
case the results would be the same, it is technically incorrect because
it assumes the two columns are paired and, in fact, they represent
completely different plots. The command should be entered as
ttest
normal
==
new,
unpaired
to be correct. Paired data would be
used where treatments are applied to the same experimental units or
you are interested in the difference between the paired values. For
example, in a study examining weight gain in steers with and without
a new wormer, weight gain may be recorded before administering the
new drug and again after administration. Here, there is a clear asso-
ciation between each pair of measurements because both are observed
in each animal.
Finally, there is an immediate form of the t-test available for both
one-sample and two-sample cases.
An example for a one-sample case would be to enter the values
ttesti
120 93.8 27.0 55
The first value is the number of observations (120), next is the test
mean (93.8), the third value is the standard deviation (27.0), and,
finally, the value (55) is the test mean with which to compare.
A NOVA
One of the most important types of analysis used is the analysis of vari-
ance (ANOVA). This expands beyond the t-test by offering a method
to analyze more than two sample means. In this type of analysis, a
specific value, called F, is calculated. This is named in honor of R. A.
Search WWH ::
Custom Search