Stats for Strategy
Fall 2008
HOMEWORK 9 (covers Topic 10, Part 2. Due Monday, Nov. 10)
DIRECTIONS:
•
Some of the exercises listed below include special instructions
which modify or clarify textbook
instructions.
•
Use a 10% significance level
for tests unless noted otherwise.
•
Data sets for all exercises are available on the
Data Sets
link from the course website.
(Remember to open data files as worksheets
in
MINITAB
.)
I. Homework Problems with Posted Solutions (A-G)
A.
Price-Fixing Litigation
•
Exercise 11.108
(p. 678)
Note:
If the actual price can be explained by supply and demand, there is
no evidence of price-fixing. (Use 95% certainty in part (a).)
B.
The Taste of Cheese
Directions:
1. Read the one-paragraph description of Case Study 1 at the bottom of page 681.
Then read more details about the
Cheese
data set on page A-1 in the Data Appendix
near the back of the book. (Assume that Taste is measured in points and that acid
concentrations in cheese are measured as percents.)
2. Make a matrix plot for all of the variables. Also get a matrix of correlations to go
along with the plots:
Stat
>
Basic Statistics
>
Correlation
>
(Select all variables)
>
OK
3. Play around with using different models (combinations of predictor variables) in
Regression
>
Regression
Observe which models have significant
t
tests at the 10% level. (There are 7 different
models which you can check.)
Answer these questions:
(a) Describe the correlations (positive or negative?) between variables as shown by the
Matrix Plot.
(b) Which predictor variable is most highly correlated with Taste? Which two predictors
are most highly correlated with each other?
(c) What does the
F
test from the
full
model indicate?
(d) Which predictor variable is ineffective when used in combination with any other
predictor? (Use 10% significance.)
1

This ** preview** has intentionally

**sections.**

*blurred***to view the full version.**

*Sign up*
(e) Which regression models (i.e. which combinations of predictors) are possible
candi-
dates
for the “best” model, as defined in the Topic 10 notes?
(f) Choose the “best” regression model. Write down the sample regression equation.
(g) Make a residuals plot for your chosen model. Does it appear that the model can be
safely used? Why or why not?
(h) Use the chosen model to interpret the slopes for each of the predictor variables in
the data set.
(i) Is Acetic related to Taste? Provide an interpretation to support your answer.
(j) Predict with 99% certainty the taste of a batch of cheese which has 5.4% Acetic,
6.9% H2S, and 1.5% Lactic.

This is the end of the preview.
Sign up
to
access the rest of the document.