Introduction to a variety of useful statistical ideas and techniques - Improving the User Experience through Practical Data Analytics - page 18

FIGURE 1.13
Excel output for inding a conidence interval.
SIDEBAR: BE CAREFUL! EXCEL'S WACKY'S NAME FOR “e”
Be aware that “conidence level” and “conidence interval” are not synonymous. Excel's “notation/
language” can be a bit confusing. The number obtained in the “conidence level” row (row 16 of
Figure 1.13 , for example) is the “error”—what above we labeled as “e.” There is no one standard
name for “e.” It is referred to as the “precision,” and also the “accuracy” (both correctly so!), or,
indeed, sometimes simply by the more general term, “error.” When your chosen conidence level is
95%, which, as we noted earlier, is most often the case—based on tradition, “e” is quite often called
“the margin of error.” This term is especially prevalent when reporting results of a poll. As indicated
in this section, when you add and subtract “e” to the X-bar—in effect straddling the mean—you've
now produced a conidence interval.
SIDEBAR: HOW TO DISPLAY CONFIDENCE INTERVALS ON A BAR CHART
IN EXCEL
As much as we'd like to hope that any audience for UX research would immediately grasp the idea of a
conidence interval, our experience tells us that it's far from a “no-brainer.” As is often the case, a pic-
ture can really help. Therefore, we always like to augment any bar chart of means (either alone or against
one another) with “error bars” to represent the conidence intervals. Here's how to do it in Excel:
1. First, calculate your different means and values of “e” or “error”; this is the value identiied in
your Excel data analysis output as “conidence level,” and the one we will add and subtract from
the mean to form a conidence interval.