Microsoft Office Tutorials and References
In Depth Information
Testing to the T
The arguments of the TTEST function are listed in Table 10-1.
Arguments of the TTEST Function
This is the reference to the range of the first array of data.
This is the reference to the range of the second array of data.
Either a 1 or 2. For a one-tailed test, enter a 1. For a 2-tailed
test, enter a 2.
Type of t-test to perform. The choice is a 1, 2, or 3. A number
1 indicates a paired test. A number 2 indicates a two-sample
test with equal variance. A number 3 indicates a two-sample
test with unequal variance.
The third argument of TTEST tells whether to conduct a one-tailed or
twotailed test. A one-tailed test is used when there is a question of whether one
set of data is specifically larger or smaller than the other. A two-tailed test is
used to tell whether the two sets are just different from each other without
specifying larger or smaller.
The first two arguments to TTEST are the ranges of the two sets of values. A
pertinent consideration here is how the two sets of data are related. The sets
could be comprised of elements that have a corresponding member in each
set. For example, there could be a set of “before” data and a set of “after” data.
Height at Week 1
Height at Week 2
3 3 ⁄ 4 inches
4 1 ⁄ 2 inches
5 1 ⁄ 2 inches
This type of data is entered into the function as paired. In other words, each
data value in the first sample is linked to a data value in the second sample.
In this case, the link is due to the fact that the data values are “before” and
“after” measurements from the same seedlings. Data can be paired in other
ways. In the salary survey, for example, each accountant may be paired with
a professor of the same age to ensure that length of time on the job does not
affect the results — in this case, you would also use a paired t-test.
When you’re using TTEST for paired samples, the two ranges entered for the
first and second arguments must be the same size. When you’re comparing
two independent (unpaired) samples, the two samples don’t have to be the