Data-Driven Business Decisions

CHAPTER 11

Are These Customers Different? Did the Intervention Work? Looking at Changes in Mean Performance

11.0. Introduction: What Is The Issue?

How do the purchasing habits of male and female customers differ? How many products do they typically purchase on a shopping trip? How much do they spend per purchase?

A telephone company mails advertising brochures to a randomly selected group of customers. How effective is the mail-out? This involves a comparison of the behavior of customers who did and did not receive the brochure.

We are interested in some performance variable X, such as the number or value of purchases per customer, customer satisfaction, or total customer waiting time in connecting to a phone consultant. Specifically, we are interested in whether or not the mean performance depends on a two-level factor like gender (male or female), day type (weekend or weekday), and strategy (advertise or don't advertise). And if it does depend on this factor, how large is the difference? And just as importantly, how reliable is our assessment?

Example 1 U.S. Labor Census revisited

In Chapter 2 we looked at the Bureau of Labor Statistics in the United States. The data described almost 8000 people who were classified as full-time workers in the previous year. The main variables recorded were salary in units of earnings per hour (EPH), educational achievement such as high school diploma (HS) or a bachelor's degree (UNI), as well as age and gender.

One issue of interest was the gender ...