Can you relate in multiple ways? Multiple linear regression and stepwise regression - Improving the User Experience through Practical Data Analytics - page 269

“Deal!” you say solemnly.
Vellucci gets up quickly, grabs his suit coat, and starts striding toward the exit. As he
passes his secretary's desk, he points back to you and Hans, still sitting on the couches.
“Dorothy, please get these folks some complimentary Red Sox tickets. They
deserve them.” He turns to you and Hans one last time before he steps into the hall.
“I know it's hard to tell, but I really do appreciate this kind of work. Thanks again!”
10.8 SUMMARY
In this chapter we introduced multiple and stepwise regression. We illustrated multiple
regression with a small data set and then applied it to the real-world prototypical problem
at Behemoth.com. That led naturally to stepwise regression, a technique that is a variation
of multiple regression, very speciically oriented toward inding the best model/equa-
tion in a world of many variables which invariably have patterns of overlap of informa-
tion about Y, the dependent variable, which are dificult to see and understand. Stepwise
regression is guaranteed to result in an equation that has only signiicant variables and
guarantees that it does not miss any variables that, if included, would be signiicant.
10.9 EXERCISE
1. Consider the Excel data in the ile “Chapter 10.Exercise 1,” which has 250
data points. The irst 12 columns represent the evaluation of a search engine on
12 respective attributes. The 12 attributes are not measured on the same scale.
Attribute 1 is a 0,1 scale (whether a respondent is able to fully fulill some tasks
using the search engine). The other Attributes are on a scale of 0-50 or 0-100,
some a function of actual results by the responder, and some a subjective evalu-
ation by the responder. Some of the attributes are “negatively scaled,” so that a
lower number is superior.
a. Run a multiple regression with “Overall Satisfaction” as the dependent
variable and attributes 1-12 as the 12 independent variables. Which attributes are
signiicant at alpha = 0.05?
b. What percent of the variability in Y is explained by all 12 attributes?
c. Using SPSS and the data ile named “Chapter 10..Exercise 1.data,” run a
stepwise regression with the default values for variables to enter and be
deleted. The output is in a ile named “Chapter 10..Exercise 1.output.”
Which attributes are in the stepwise regression model at the last step?
What percent of the variability in Y is explained by the variables that are
in the stepwise regression model at the last step?
The answers are in a Word ile named “Chapter 10.Exercise 1. answers.”