REPORTING THE RESULTS 1 15the intervals. When using t, it may be difficult to find tabled values close to those
you want. The simplest thing to do is to use a computer program that gives t values
for any number between 0 and 1. As a rough approximation, the linear interpolation
given in Section 6.2.2 may be used. Other commonly used multiple comparison tests
for testing equality of means include the Tukey HSD (honestly significant difference),
Scheffk’s test or Fisher’s LSD test (see Milliken and Johnson [1984]).
8.8 REPORTING THE RESULTS
In typical biomedical examples involving patients, numerous variables are measured.
A few of these variables relate to the main purpose of the study and are called outcome
variables. Here, tests of hypotheses are commonly reported. But other variables are
collected that describe the individual patients, such as age, gender, seriousness of
medical condition, results from a wide range of medical tests, attitudes, and previous
treatment. These data are taken so that the patients being treated can be adequately
described and to see if the patients in the various treatment groups were similar prior
to treatment. In experiments using animals and in laboratory experiments, usually
fewer variables are measured. In reporting the results, we have to decide what is of
interest to the reader.
When we report the results of tests of hypotheses, we should include not only the
P value but also other information from the sample that will aid in interpreting the
results. Simply reporting the P value is not sufficient. It is common when reporting
the results of HO : p1 = p2 to also report the two sample means so that the reader can
see their actual values. Usually, either the standard deviations or the standard errors
of the means are also included. The standard deviation is given if we want the reader
to have a measure of the variation of the observations and the standard error of the
mean is given if it is more important to know the variation of the mean value. The
standard deviation is sometimes easier to interpret than the standard error of the mean
if the sample sizes in the two groups are unequal since its size does not depend on the
sample size. Ranges are useful in warning the reader that outliers may be present.
Often tables of means and standard deviations are given for variables that simply
describe the patients in the study. This information is often given in a graphical form
with bar graphs. If the shape of the distribution is of interest, then using histograms,
box plots, or other available options are useful. These plots can be obtained directly
from the statistical programs, and many readers find them easier to interpret than
tables of statistics.
PROBLEMSIn the following problems, use the steps listed in Section 8.1.3 as a guideline when
performing tests. Most of these test can be done using computer programs.
8.1 In Problem 7.5, make a test to determine whether or not pressure increases
mean bleeding time. Use cy = .05.
In Problem 7.4, test whether male rats are heavier, on average, than female rats.
Use cy = .05.8.2