Exam Marketing Research Methods
Thursday, April 7 2016
Teacher
The exam contains 23 pages.
The exam consists of 3 parts with several sub-questions.
Please answer each question in the corresponding answer box.
You have 3 hours to complete the exam (8.30-11.30 hrs).
Make sure that you write your name and student number the first page!
Please answer in English (Dutch is not allowed by the faculty).
The grades for the exam will be published on Nestor.
We wish you lots of success!!!
Name
Student number
Year / semester you followed MRM
Maximum number of credits per question:
Part 1: AN(C)OVA, Regression, and Third variables 36 credits
Part 2: Conjoint analysis & Binary Data 24 credits
Part 3: Factor, Reliability & Cluster analysis 40 credits
1
,Part 1. AN(C)OVA, Regression, and Third variables (24 credits)
1- In order to analyze the effect of advertisement on sales, a marketing analyst generates a factorial
variable with the following levels:
Level of advertisement advFACT Description
0 0 No advertisement
Between 0 and 200000 Euros 1 Moderate advertisement
Larger than 200000 Euros 2 High advertisement
Then, she looks for difference in level of sales for these three different levels of advertisement. The
ANOVA table below shows the results of this analysis.
ANOVA
vsal
Sum of Squares df Mean Square F Sig.
Between Groups 2870045054954,479 2 1435022527477,239 7,966 ,000
Within Groups 36748587663393,350 204 180140135604,869
Total 39618632718347,830 206
a. What does the analysis above tell us about the effect of promotional activities? (3 credits )
Variable advFACT has three levels, so the analysis above compares the average sales for three cases. The F-
statistic stating the difference in variances is insignificant, and hence there is significant difference between
the sales when there is no ad, moderate ad and high ad.
Now we introduce price as covariate. The following ANCOVA tables pop up.
Between-Subjects Factors
N
advFACT ,00 81
1,00 93
2
, 2,00 33
Tests of Between-Subjects Effects
Dependent Variable: vsal
Type III Sum of
Source Squares df Mean Square F Sig.
a
Corrected Model 6858109265560,531 4 1714527316390,133 10,572 ,000
Intercept 10289606034459,162 1 10289606034459,162 63,445 ,000
advFACTprice 476443880051,719 1 476443880051,719 2,938 ,088
price 580051052554,199 1 580051052554,199 3,577 ,060
advFACT 1300848918516,524 2 650424459258,262 4,010 ,020
Error 32760523452787,273 202 162180809172,214
Total 217069386280800,000 207
Corrected Total 39618632718347,805 206
a. R Squared = ,173 (Adjusted R Squared = ,157)
b. What does the test for homogeneity of slopes say? Why is this test important? Can we
continue with our ANCOVA analysis? (5 credits )
Homogeneity of slopes means that the covariate has the same effect on the dv for all levels of the factorial
variable. To test for that the interaction term of the covariate and the factorial variable has to be included
as an additional covariate. By the interaction not being significant, we assure the homogeneity of slopes.
This is important because we assume the same ratio of variance of the dv on covariate across all the levels
of factorial variable, if that is not the case, the ANCOVA’s results are not reliable.
Here we have significance level of .06 which is marginally larger than .05, so we can continue, however with
some doubts…
3
Thursday, April 7 2016
Teacher
The exam contains 23 pages.
The exam consists of 3 parts with several sub-questions.
Please answer each question in the corresponding answer box.
You have 3 hours to complete the exam (8.30-11.30 hrs).
Make sure that you write your name and student number the first page!
Please answer in English (Dutch is not allowed by the faculty).
The grades for the exam will be published on Nestor.
We wish you lots of success!!!
Name
Student number
Year / semester you followed MRM
Maximum number of credits per question:
Part 1: AN(C)OVA, Regression, and Third variables 36 credits
Part 2: Conjoint analysis & Binary Data 24 credits
Part 3: Factor, Reliability & Cluster analysis 40 credits
1
,Part 1. AN(C)OVA, Regression, and Third variables (24 credits)
1- In order to analyze the effect of advertisement on sales, a marketing analyst generates a factorial
variable with the following levels:
Level of advertisement advFACT Description
0 0 No advertisement
Between 0 and 200000 Euros 1 Moderate advertisement
Larger than 200000 Euros 2 High advertisement
Then, she looks for difference in level of sales for these three different levels of advertisement. The
ANOVA table below shows the results of this analysis.
ANOVA
vsal
Sum of Squares df Mean Square F Sig.
Between Groups 2870045054954,479 2 1435022527477,239 7,966 ,000
Within Groups 36748587663393,350 204 180140135604,869
Total 39618632718347,830 206
a. What does the analysis above tell us about the effect of promotional activities? (3 credits )
Variable advFACT has three levels, so the analysis above compares the average sales for three cases. The F-
statistic stating the difference in variances is insignificant, and hence there is significant difference between
the sales when there is no ad, moderate ad and high ad.
Now we introduce price as covariate. The following ANCOVA tables pop up.
Between-Subjects Factors
N
advFACT ,00 81
1,00 93
2
, 2,00 33
Tests of Between-Subjects Effects
Dependent Variable: vsal
Type III Sum of
Source Squares df Mean Square F Sig.
a
Corrected Model 6858109265560,531 4 1714527316390,133 10,572 ,000
Intercept 10289606034459,162 1 10289606034459,162 63,445 ,000
advFACTprice 476443880051,719 1 476443880051,719 2,938 ,088
price 580051052554,199 1 580051052554,199 3,577 ,060
advFACT 1300848918516,524 2 650424459258,262 4,010 ,020
Error 32760523452787,273 202 162180809172,214
Total 217069386280800,000 207
Corrected Total 39618632718347,805 206
a. R Squared = ,173 (Adjusted R Squared = ,157)
b. What does the test for homogeneity of slopes say? Why is this test important? Can we
continue with our ANCOVA analysis? (5 credits )
Homogeneity of slopes means that the covariate has the same effect on the dv for all levels of the factorial
variable. To test for that the interaction term of the covariate and the factorial variable has to be included
as an additional covariate. By the interaction not being significant, we assure the homogeneity of slopes.
This is important because we assume the same ratio of variance of the dv on covariate across all the levels
of factorial variable, if that is not the case, the ANCOVA’s results are not reliable.
Here we have significance level of .06 which is marginally larger than .05, so we can continue, however with
some doubts…
3
