So you take statistics, and you know you need to use
a t-test, but are confused about what kind of t-test
to use? This simple article will show you how you
can determine if a paired, unpaired, or one-sample
t-test is appropriate in your particular situation.
1. Ask yourself: do I want to compare the means of
two groups, or should I just don't care about how
the average of a single group are compared with some
numbers? If you want to compare the means of two
groups, continue to step 2.
But if you only care how the average of a single
group in relation to a single number, you must use a
one-sample t-test. An example of a case where a
one-sample t-test is appropriate would be if you are
testing whether the average student spends
considerably more than 2000 calories a day (e.g.
Comparing the average number of calories consumed in
order to see if it's significantly greater than the
number of 2000).
2. If you compare means of two groups, the next ask
yourself: Have the two groups of figures which we
compare comes from the same people? If so, we need
to use a paired samples t-test (also known as a
repeated-samples t-test).
For example, let's say that we compare the weight of
each person in a group of people before they went on
a diet with their weight after they completed diet
program. We would like to know whether each person's
weight when the program is significantly greater
than their weight in advance. The two sets of
numbers, we compare come from the same set of
people: a set represents their weight before
treatment, and the second set represents their
weight after treatment. This is called a
within-subjects variable. In a case like this, you
must use a paired samples t-test (also known as a
repeated-samples t-test).
There is even a case where a paired samples t-test
is appropriate. : If the researcher doing a
"matched" design, where the deliberately chosen few
of the items that are the same for different
characteristics (e.g., age, gender, medical history,
etc.) Anytime that the figures in the first and
second group are paired, there is a meaningful link
between a value in the first group of scores and the
corresponding value in the other group of scores, a
paired samples t-test is appropriate.
3. In all other cases where a t-test is appropriate,
it is best to use an independent samples t-test.
This is appropriate for the "between-subjects"
designs, where two groups of subjects are meant to
differ on a critical manipulation. For example, if
testing the effect of caffeine on the growth of
plants, you can have two groups: a control group,
who were given water, and an experimental group of
plants, which got a caffeine solution. When you use
completely different plants in each group, there is
no meaningful mating between the scores in the two
groups, and you must use an independent-samples
t-test. |