Comparing two designs (or anything else!) using paired sample T-tests - Improving the User Experience through Practical Data Analytics - page 78

the task using either form type), while some of the participants were somewhat slow
at the task (e.g., participants 3 and 9, who required a relatively long time to perform
the task using each of the designs). We point this out, since it is related to why there
is a difference in the speciic way the t-test is carried out by the software, depending
on whether the user indicates that it is an “independent t-test”—(as illustrated in the
previous chapter) or a “paired t-test”—being illustrated in this chapter.
3.4.1 EXCEL
The means for the two designs are 12.88 minutes (“The Long Scroller”) and
14.26 minutes (“The Wizard”). But, as we know from hypothesis testing logic, we
do not yet know if this difference in sample means is indicative of a true difference
in means. Of course, we might have a strong intuitive suspicion that the difference of
nearly 1.5 minutes will indicate that there is a real difference; that the difference is
large enough to reject H0. Perhaps; we'll see!
We enter the data as shown, and we open “Data Analysis” in Excel as depicted
in Figure 3.1 .
Note that the command “t-Test: Paired Two-Sample for Means” (see arrow in
Figure 3.1 ) is directly above the commands examined in the previous chapter on
t-tests for independent samples.
FIGURE 3.1
Opening data analysis to perform a t-test: paired two sample for means; Excel.