Introduction to a variety of useful statistical ideas and techniques - Improving the User Experience through Practical Data Analytics - page 4

time to complete a task, etc.) between the two points. The easiest way to do this is to
use an Excel command—a command that is already built into Excel and is easy to
use. But, before we get into the calculations, let's provide a scenario that will help
you see the value of doing the calculations in the irst place.
1.2.1.1 Vignette: how long does it take to hook up DSL Internet service?
You're the UX researcher at a gigantic telecommunications company that provides
broadband service to hundreds of thousands of customers in the United States. You're
part of the user experience team that is dedicated to improving the user experience
for those broadband clients. One of the key projects you're working on is trying to
decrease the time it takes for a typical new customer to install DSL Internet ser-
vice. (In the past the company sent out technicians to complete the installation, but
has recently gone to a “customer self-serve” model of sending out DSL installation
kits. The cost savings have been dramatic.) Shorter installation times usually indicate
higher satisfaction, and fewer calls to technical support, which means signiicant cost
savings. During home visits to watch folks installing DSL, you take many measure-
ments—time to download necessary software, time to activate the account, etc.
But one of the most important metrics is simply the time to hook everything
up. That is, plugging a phone line to the modem, plugging the modem to the com-
puter, iltering the existing phone line, etc. Obviously, the shorter the better for the
customer, but for the company as well, because it means fewer calls to the customer
support center in Mumbai, India—they're already swamped.
Your collected data indicates that the average time, μ, to hook everything up is 160
seconds (2 minutes, 40 seconds), and that the variability factor (which, remember, is
a measure of how different the time required is from person to person), is represented
by a standard deviation, σ, of 20 seconds. You know that the “hook up” time is viewed
historically as around 180 seconds (3 minutes) and, to beef up the report—and antici-
pating the question during your presentation—you decide to calculate the percentage
of the users who will require less than 190 seconds to hook everything up.
OK, so we have the average task completion time value (μ) = 160 seconds with a
standard deviation (σ) of 20 seconds. And we assume that we have a set of data that
follows a normal curve (although not every process in the world follows a normal
curve; more about the prevalence of the normal curve later), and a picture of what we
want is the shaded-in area of the normal curve in Figure 1.4 .
The letter, X, here represents the time to hook everything up. In essence, X is a
traditional statistical symbol; statisticians would refer to it as a “random variable.”
We would now ind the shaded-in area (i.e., the probability that the time to hook
everything up is less than 190 seconds)—or percentage 1 —by using the following
Excel command:
=NORMDIST(X , μ , σ , 1),
1 Actually, we need to multiply the value by 100 and add a percent sign to get the answer as a percent.
The command provides the value from 0 to 1, in essence, a probability or a proportion.