Statistics for Managers
Using Microsoft® Excel
5th Edition
Chapter 15
Multiple Regression Model Building
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-1
Learning Objectives
In this chapter, you learn:
To use quadratic terms in a regression model
To use transformed variables in a regression model
To measure the correlation among independent
variables
To build a regression model, using either the
stepwise or best-subsets approach
To avoid the pitfalls involved in developing a
multiple regression model
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-2
Nonlinear Relationships
The relationship between the dependent variable
and an independent variable may not be linear
Can review the scatter plot to check for non-
linear relationships
Example: Quadratic model
Yi β 0 β1X1i β 2 X1i2 ε i
The second independent variable is the square of the
first variable
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-3
Quadratic Regression Model
Model form:
Yi β0 β1X1i β2 X ε i 2
1i
where:
β0 = Y intercept
β1 = regression coefficient for linear effect of X on Y
β2 = regression coefficient for quadratic effect on Y
εi = random error in Y for observation i
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-4
Quadratic Regression Model
Y Y
X X
residuals
X residuals X
Nonlinear not does not Linear fit does give random
give random residuals residuals
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-5
Quadratic Regression Model
Yi β0 β1X1i β 2 X1i2 ε i
Quadratic models may be considered when the scatter
diagram takes on one of the following shapes:
Y Y Y Y
X1 X1 X1 X1
β1 < 0 β1 > 0 β1 < 0 β1 > 0
β2 > 0 β2 > 0 β2 < 0 β2 < 0
β1 = the coefficient of the linear term
β2 = the coefficient of the squared term
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-6
Quadratic Regression
Equation
Collect data and use Excel to calculate the
regression equation:
Ŷi b0 b1X1i b 2 X 2
1i
Test for Overall Relationship
H0: β1 = β2 = 0 (no overall relationship between X and Y)
H1: β1 and/or β2 ≠ 0 (there is a relationship between X and Y)
MSR
F-test statistic = MSE
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-7
Testing for Significance
Quadratic Effect
Testing the Quadratic Effect
Compare quadratic regression equation
^
Yi b 0 b1X1i b 2 X1i2
with the linear regression equation
^
Yi b 0 b1X1i
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-8
Testing for Significance
Quadratic Effect
Testing the Quadratic Effect
Consider the quadratic regression equation
^
Yi b 0 b1X1i b 2 X1i2
Hypotheses
H0 : β 2 = 0 (The quadratic term does not improve the
model)
H1 : β 2 0
(The quadratic term improves the model)
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-9
Testing for Significance
Quadratic Effect
Testing the Quadratic Effect
Hypotheses
H0: β2 = 0 (The quadratic term does not improve the model)
H1: β2 0 (The quadratic term improves the model)
The test statistic is
b2 β2 where:
t
Sb 2 b2 = estimated slope
β2 = hypothesized slope (zero)
d.f. n 3 Sb2 = standard error of the slope
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-10
Testing for Significance
Quadratic Effect
Testing the Quadratic Effect
If the t test for the quadratic effect is
significant, keep the quadratic term in the
model.
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-11
Quadratic Regression
Example
Filter Purity increases as filter time increases:
Purity Time
3 1
Purity vs. Time
7 2
8 3 100
15 5
22 7 80
33 8
60
P urity
40 10
54 12
40
67 13
70 14 20
78 15
85 15 0
87 16 0 5 10 15 20
99 17 Time
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-12
Quadratic Regression
Example
Simple (linear) regression results:
^
Y = -11.283 + 5.985 Time t statistic, F statistic, and
2
adjusted r are all high, but
Standard the residuals are not random:
Coefficients Error t Stat P-value
Intercept -11.28267 3.46805 -3.25332 0.00691
Time Residual Plot
Time 5.98520 0.30966 19.32819 2.078E-10
10
Regression Statistics 5
F Significance F
R Square 0.96888
0
esiduals
Adjusted R Square 0.96628 373.57904 2.0778E-10
-5 0 5 10 15 20
R
Standard Error 6.15997
-10
Time
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-13
Quadratic Regression
Example
Quadratic regression results:
Time Residual Plot
^
Y = 1.539 + 1.565 Time + 0.245 (Time)2 10
Residuals
Standard
Coefficients Error t Stat P-value 0
Intercept 1.53870 2.24465 0.68550 0.50722 0 5 10 15 20
-5
Time 1.56496 0.60179 2.60052 0.02467 Time
Time-squared 0.24516 0.03258 7.52406 1.165E-05
Time -square d Residual Plot
Regression Statistics F Significance F 10
R Square 0.99494 1080.7330 2.368E-13 5
Residuals
Adjusted R Square 0.99402
0
Standard Error 2.59513 0 100 200 300 400
-5
The quadratic term is significant and improves Time-squared
the model: adjusted r2 is higher and SYX is
lower, residuals are now random.
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-14
Using Transformations in
Regression Analysis
Idea:
Non-linear models can often be transformed to a
linear form
Can be estimated by least squares if
transformed
Transform X or Y or both to get a better fit or to
deal with violations of regression assumptions
Can be based on theory, logic or scatter plots
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-15
The Square Root
Transformation
The square-root transformation
Yi β0 β1 X1i ε i
Used to
overcome violations of the equal variance assumption
fit a non-linear relationship
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-16
The Square Root
Transformation
Yi β0 β1X1i ε i Yi β0 β1 X1i ε i
Shape of original relationship Relationship when transformed
Y Y b1 > 0
X X
Y Y
b1 < 0
X X
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-17
The Log Transformation
The Multiplicative Model:
Original multiplicative model Transformed multiplicative model
Yi β0 X1iβ1 ε i log Yi log β0 β1 log X1i log ε i
The Exponential Model:
Original multiplicative model Transformed exponential model
Yi e
β0 β1X1i β 2 X 2i
εi ln Yi β0 β1X1i β 2 X 2i ln ε i
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-18
Interpretation of Coefficients
For the multiplicative model:
log Yi log β0 β1 log X1i log ε i
When both dependent and independent variables are
transformed:
The coefficient of the independent variable Xk can be
interpreted as follows: a 1 percent change in Xk leads
to an estimated bk percentage change in the mean value
of Y.
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-19
Collinearity
Collinearity: High correlation exists among
two or more independent variables
The correlated variables contribute redundant
information to the multiple regression model
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-20
Collinearity
Including two highly correlated independent
variables can adversely affect the regression
results
No new information provided
Can lead to unstable coefficients (large
standard error and low t-values)
Coefficient signs may not match prior
expectations
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-21
Some Indications of Strong
Collinearity
Incorrect signs on the coefficients
Large change in the value of a previous coefficient
when a new variable is added to the model
A previously significant variable becomes
insignificant when a new independent variable is
added
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-22
Detecting Collinearity
Variance Inflationary Factor
VIFj is used to measure collinearity:
1
VIFj
1 R j
2
where R2j is the coefficient of determination from a regression
model that uses Xj as the dependent variable and all other X
variables as the independent variables
If VIFj > 5, Xj is highly correlated with the
other independent variables
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-23
Model Building
Goal is to develop a model with the best set of
independent variables
Easier to interpret if unimportant variables are removed
Lower probability of collinearity
Stepwise regression procedure
Provide evaluation of alternative models as variables are
added
Best-subset approach
Try all combinations and select the model with the
highest adjusted r2
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-24
Stepwise Regression
Idea: develop the least squares regression
equation in steps, adding one independent
variable at a time and evaluating whether
existing variables should remain or be
removed
The coefficient of partial determination is the
measure of the marginal contribution of each
independent variable, given that other
independent variables are in the model
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-25
Best Subsets Regression
Idea: estimate all possible regression equations using all
possible combinations of independent variables
Choose the best model by looking for the highest adjusted r2
The model with the largest adjusted r2 will also have the
smallest SYX
Stepwise regression and best subsets regression can
be performed using Excel with PHStat add in
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-26
Alternative Best Subsets
Criterion
Calculate the value Cp for each potential
regression model
Consider models with Cp values close to or
below k + 1
k is the number of independent variables in the
model under consideration
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-27
Alternative Best Subsets
Criterion
The Cp Statistic
(1 Rk2 )(n T )
Cp (n 2(k 1))
1 RT2
Where k = number of independent variables included in a
particular regression model
T = total number of parameters to be estimated in the
full regression model
Rk2 = coefficient of multiple determination for model with k
independent variables
R 2T = coefficient of multiple determination for full model with
all T estimated parameters
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-28
8 Steps in Model Building
1. Choose independent variables to include in the model
2. Estimate full model and check VIFs and check if any VIFs > 5
If no VIF > 5, go to step 3
If one VIF > 5, remove this variable
If more than one, eliminate the variable with the highest VIF
and repeat step 2
3. Perform best subsets regression with remaining variables.
4. List all models with Cp close to or less than (k + 1).
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-29
8 Steps in Model Building
5. Choose the best model.
Consider parsimony.
Do extra variables make a significant contribution?
6. Perform complete analysis with chosen model, including
residual analysis.
7. Transform the model if necessary to deal with violations of
linearity or other model assumptions.
8. Use the model for prediction and inference.
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-30
Model Validation
The final step in the model-building process is
to validate the selected regression model.
Collect new data and compare the results.
Compare the results of the regression model
to previous results.
If the data set is large, split the data into two
parts and cross-validate the results.
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-31
Model Building Flowchart
Choose X1,X2,…,Xk Run best subsets
regression to obtain
No “best” models in terms
Run regression to Any of Cp
find VIFs VIF>5?
Yes Do complete analysis
Remove Yes
variable with More than
Add quadratic term and/or
highest one?
transform variables as indicated
VIF
No
Remove this X Perform predictions
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-32
Pitfalls and Ethical
Considerations
To avoid pitfalls and address ethical considerations:
Understand that interpretation of the estimated
regression coefficients are performed holding all
other independent variables constant
Evaluate residual plots for each independent
variable
Evaluate interaction terms
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-33
Pitfalls and Ethical
Considerations
To avoid pitfalls and address ethical considerations:
Obtain VIFs for each independent variable before
determining which variables should be included in the
model.
Examine several alternative models using best-subsets
regression.
Use other methods when the assumptions necessary for
least-squares regression have been seriously violated.
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-34
Chapter Summary
In this chapter, we have
Developed the quadratic regression model
Discussed using transformations in regression
models
The multiplicative model
The exponential model
Described collinearity
Discussed model building
Stepwise regression
Best subsets
Addressed pitfalls in multiple regression and ethical
considerations
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc. Chap 15-35