=========================preview======================
(mark222)[2008](s)exam2~PPSpider^sol_10375.pdf
Back to MARK222 Login to download
======================================================
MARK 222 Marketing Research
Spring 2008
Exam II
Name (in block letter): Answer Key
Student ID:
Section Number:
Total 200 points. 1 hour and 20 minute exam. Be concise and clear in writing answers.
There are 7 pages including this page. Make sure that your copy has all 7 pages.
Part I. 20 points for each
Suppose the following questionnaire is used to analyze the importance of attributes.
Q1. Allocate 100 points among the following attributes according to the importance weight when buying a soft drink.
Taste ( )
Price ( )
Brand ( )
Trend ( )
Packaging ( )
Nutrition Value ( )
Calories Level ( )
Sum 100
Q2. What is your occupation?
Student( ) Housewife( ) Employed( ) Self-employed( ) Others( )
1. Explain how to code the questionnaire. (That is, make a code book.)
For Q1, define 7 ratio scaled variables. Input the number from the answer directly. For Q2, define a nominal scaled variable. 1-student, 2-housewife, 3-employed, 4-self-employed, 5-others.
2. Suppose we collect opinions from 1000 consumers. Explain how to compare the relative importance of Price over Nutrition Value at the market level.
Conduct the mean difference testing using t-test. H0: mean of price=mean of nutrition.
3. Explain what analytical method can be used to check if consumers with different occupations have different importance weight on Price . What is the basic assumption of such analytical method in this particular case?
Since there are 5 groups in the occupation dimension, we cant use t-test. Instead, we use ANOVA. The basic assumption is that the within-group variance is the same for all groups. In this case, the variance of importance weight on price is the same for all different type of occupations.
Part II. Regression Model, 25 points each
1. We know from marketing theory that price has a negative impact on sales while advertising has a positive impact on sales. Suppose price and advertising variables are positively correlated. Would omitting the advertising variable from the regression model result in exaggerating the effect of price on sales?
No. There would be a upward bias in the parameter estimate. Specifically, the expected parameter estimate in the omitted model would be Cov(sales, price)/Var(price)=Cov(a+b1*price+b2*ad, price)/Var(price) = b1+b2*Cov(ad,price)/Var(price)>b1 since b2>0 and Cov(ad, price)>0. That is, the estimated parameter in the omitted model would be less negative than the true parameter. So the effect of price reduction on demand would be estimated to be smaller in magnitude than the true level. Thus, it would result in the underestimation of price effect.
2. Consider a regression model to explain the variation in the satisfaction toward the university among the third year students in UST. The dependent variable is a seven-point rating of satisfaction. There are three factors that will be used to explain the satisfaction score: students m