Exercise 3

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Exercise 3

• Question 1 will be covered in the tutorial


• Question 2, 3, 4 and 5 are exercises to be completed on your own. You may submit them
to the TA if you want feedback (this has no impact on your final grades).
• WE STRONGLY RECOMMEND TRYING QUESTION 4
• IF YOU’RE SHORT ON TIME, SKIP QUESTION 5. IT’S A GUIDED EXAMPLE OF A
GROUP PROJECT

Question 1
For this question, we use the dataset music.sav from question 2 from the past week. The variables are
as follows:
• Store_nr: id for the store
• Music: 1 if no music is played in the store, 2 is music with low volume is played, and 3 if music with high volume
is played in the store
• Sales: Store sales in Euros
• Size: size of the store in square meter

When answering the following questions, you may always assume ceteris paribus.
a) Create a new variable Instoremusic that takes value 1 if music is played in store and 0
otherwise
b) Estimate the following model, and interpret the resulting coefficients. Clearly state whether the
impact of either variable is significant. What is the effect of playing music in stores on sales?
Also interpret the model fit.
𝑆𝑎𝑙𝑒𝑠 = 𝛽0 + 𝛽1 𝑖𝑛𝑠𝑡𝑜𝑟𝑒𝑚𝑢𝑠𝑖𝑐 + 𝛽2 𝑠𝑖𝑧𝑒
c) Now, estimate the following model, and interpret the resulting coefficients. Discuss differences
with b)
𝑆𝑎𝑙𝑒𝑠 = 𝛽0 + 𝛽1 𝑛𝑜𝑚𝑢𝑠𝑖𝑐 + 𝛽2 𝑙𝑜𝑤𝑚𝑢𝑠𝑖𝑐 + 𝛽3 𝑠𝑖𝑧𝑒
where:
• nomusic= 1 if no music is played in store, 0 otherwise
• lowmusic= 1 if music with low volume is played in store, 0 otherwise
What is the expected difference in sales from playing low music versus playing high music in
stores?
d) Can you estimate the following model? If not, why?
𝑆𝑎𝑙𝑒𝑠 = 𝛽0 + 𝛽1 𝑛𝑜𝑚𝑢𝑠𝑖𝑐 + 𝛽2 𝑙𝑜𝑤𝑚𝑢𝑠𝑖𝑐 + 𝛽3 ℎ𝑖𝑔ℎ𝑚𝑢𝑠𝑖𝑐 + 𝛽4 𝑠𝑖𝑧𝑒
• highmusic= 1 if music with high volume is played in store, 0 otherwise

e) Estimate the following model and answer the questions below:


log (𝑆𝑎𝑙𝑒𝑠) = 𝛽0 + 𝛽1 𝑛𝑜𝑚𝑢𝑠𝑖𝑐 + 𝛽2 𝑙𝑜𝑤𝑚𝑢𝑠𝑖𝑐 + 𝛽3 log (𝑠𝑖𝑧𝑒)
o What is the elasticity of sales with respect to store size?
o What are the expected sales with low music if the sales with high music are 10000
euros?
o What is the expected percentage difference in sales between stores with low music
versus stores with high music?
o Which variable has the highest (positive) effect on sales?

Question 2
We reconsider study 5 from the article "Persuasion, interrupted: The effect of momentary interruptions
on message processing and persuasion" by Kupor and Tormala (2015). We have already analyzed this
study in the past week using ANOVA. This week we use linear regression. Download the dataset used
in week 2: KT2study5Qs3.xlsx

a) Create a new dummy variable for the interaction effect between “Momentary interruption” and
“Weak argument”. Then, estimate the following model:
𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒 = 𝛽0 + 𝛽1 𝑀𝑜𝑚𝑒𝑛𝑡𝑎𝑟𝑦 + 𝛽2 𝑊𝑒𝑎𝑘 + 𝛽3 𝑀𝑜𝑚𝑒𝑛𝑡𝑎𝑟𝑦 ∗ 𝑊𝑒𝑎𝑘
b) Interpret all coefficients: which effects are significant? In particular:
o What is the expected attitude of a participant who is momentarily interrupted when the
argument is weak?
o What is the expected attitude of a participant who is continuously interrupted when the
argument is strong?
c) Compare the results with the ANOVA output you obtained in exercise from the past week.
Comment on similarity and differences.

Question 3
You are provided with a data set that contains weekly sales, prices and promotion of a popular cola
brand (let’s denote it as C-Cola) and its competitors in one grocery store. The data set contains the
following variables.

• Sales: Unit sales of C-Cola in liters


• Own Price: Unit price of C-Cola in €/liter
• Own Display: 1 if C-Cola is displayed in end-of-aisle or at checkout counters, 0 otherwise
• Own Feature: 1 if C-Cola appeared in store flyer, 0 otherwise
• Comp Price: Average price of other cola brands (in €/liter)
• Comp Disp: 1 if one of the other cola brands is displayed in end-of-aisle or at checkout counters, 0 otherwise
• Comp Feature: 1 if one of the other cola brands is featured in the store flyer, 0 otherwise

Task
Estimate the following linear regression model. You do not have to test for the assumptions of the
linear regression model.

(1) ln(𝑆𝑎𝑙𝑒𝑠𝑡 ) = 𝛽0 + 𝛽1 ln(𝑂𝑤𝑛𝑃𝑟𝑖𝑐𝑒𝑡 ) + 𝛽2 ln(𝐶𝑜𝑚𝑝𝑃𝑟𝑖𝑐𝑒𝑡 ) + 𝛽3 𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 +


𝛽4 𝑂𝑤𝑛𝐹𝑒𝑎𝑡𝑢𝑟𝑒𝑡 + 𝛽5 𝐶𝑜𝑚𝑝𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 + 𝛽6 𝐶𝑜𝑚𝑝𝐹𝑒𝑎𝑡𝑢𝑟𝑒𝑡 + 𝜀𝑡

where 𝑙𝑛 denotes logarithm (base e), t denotes the week and 𝜀𝑡 denotes the error term.

As usual, assume ceteris paribus


1. What is the own price elasticity of demand for C-Cola? In other words, what % will the sales of C-
Cola change if the price of C-Cola is increased by 1%?
2. What is the competitive price elasticity of demand for C-Cola? In other words, what % will the
sales of C-Cola change if the average price of competitors is increased by 1%?

To answer Q3-Q7, consider this example question. What will be the sales (in liters) of C-Cola when it
is displayed in end-of-aisle or at checkout counters if the sales without display is 1000 liters?

When 𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 0,
ln(𝑆𝑎𝑙𝑒𝑠𝑡 ) |(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 0)
= 𝛽0 + 𝛽1 ln(𝑂𝑤𝑛𝑃𝑟𝑖𝑐𝑒𝑡 ) + 𝛽2 ln(𝐶𝑜𝑚𝑝𝑃𝑟𝑖𝑐𝑒𝑡 ) + 𝛽4 𝑂𝑤𝑛𝐹𝑒𝑎𝑡𝑢𝑟𝑒𝑡
+ 𝛽5 𝐶𝑜𝑚𝑝𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 + 𝛽6 𝐶𝑜𝑚𝑝𝐹𝑒𝑎𝑡𝑢𝑟𝑒𝑡

When 𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 1,

ln(𝑆𝑎𝑙𝑒𝑠𝑡 ) |(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 1)
= 𝛽0 + 𝛽1 ln(𝑂𝑤𝑛𝑃𝑟𝑖𝑐𝑒𝑡 ) + 𝛽2 ln(𝐶𝑜𝑚𝑝𝑃𝑟𝑖𝑐𝑒𝑡 ) + 𝛽3 + 𝛽4 𝑂𝑤𝑛𝐹𝑒𝑎𝑡𝑢𝑟𝑒𝑡
+ 𝛽5 𝐶𝑜𝑚𝑝𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 + 𝛽6 𝐶𝑜𝑚𝑝𝐹𝑒𝑎𝑡𝑢𝑟𝑒𝑡

Difference in log sales = ln(𝑆𝑎𝑙𝑒𝑠𝑡 ) |(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 1) − ln(𝑆𝑎𝑙𝑒𝑠𝑡 )|(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 0) = 𝛽3

𝑆𝑎𝑙𝑒𝑠 |(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦 =1) 𝑆𝑎𝑙𝑒𝑠 |(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦 =1)


⇒ ln (𝑆𝑎𝑙𝑒𝑠𝑡|(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡=0)) = 𝛽3 ⇒ 𝑆𝑎𝑙𝑒𝑠𝑡|(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡=0) = 𝑒 𝛽3
𝑡 𝑡 𝑡 𝑡

⇒ 𝑆𝑎𝑙𝑒𝑠𝑡 |(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 1) = 𝑒 𝛽3 × 𝑆𝑎𝑙𝑒𝑠𝑡 |(𝑂𝑤𝑛𝐷𝑖𝑠𝑝𝑙𝑎𝑦𝑡 = 0) = 1000𝑒 𝛽3

3. What will be the expected sales (in liters) of C-Cola when it is displayed in end-of-aisle or at
checkout counters if the sales when C-Cola is not displayed is 1000 liters?
4. What will be the expected sales (in liters) of C-Cola when the competitor’s product is displayed in
end-of-aisle or at checkout counters if the sales of C-Cola without competitor’s product being
displayed is 1000 liters?
5. What will be the expected sales (in liters) of C-Cola when it is featured in the flyer if the sales of
C-Cola without it being featured in the flyer is 1000 liters?
6. What will be the expected sales (in liters) of C-Cola when the competitor’s product is featured in
the flyer if the sales of C-Cola without competitor’s product being featured is 1000 liters?
7. Among the independent variables, which of them has the highest impact on C-Cola’s sales, and
which of them has the lowest but significant impact on C-Cola’s sales?

Question 4 (adapted from Exam October 2019)

In a follow up study to Kupor and Tormala (2015), we investigate the impact of various variables on
the consumer attitude towards a novel product. Respondents are shown an advertisement, and their
attitudes are manipulated across multiple variables. We consider the following variables:
• Female: 1 if respondent is female, 0 otherwise
• Interruption: 1 if not interrupted, 2 if interrupted at the beginning of the advertisement, 3 if
interrupted in the middle of the advertisement, 4 if interrupted at the end of the advertisement
• Music: 1 if the advertisement has a musical soundtrack, 0 otherwise
• Color: 1 if advertisement has bright, flashy colors, 0 otherwise

We consider 3 possible models:


Model 1: 𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑖 = 𝛽0 + 𝛽1 Female𝑖 + 𝛽2 Music𝑖 + 𝛽3 (Music𝑖 × Female𝑖 )
For which the following SPSS linear regression output was obtained:

ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 598,331 3 199,444 223,518 ,000b
Residual 264,119 296 ,892
Total 862,451 299
a. Dependent Variable: Attitude
b. Predictors: (Constant), Female, Music, Female*Music

Coefficientsa
Standardized
Unstandardized Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 1,973 ,094 20,883 ,000
Female 1,993 ,134 ,588 14,915 ,000
Music ,895 ,164 ,249 5,471 ,000
Female*Music 1,184 ,231 ,260 5,119 ,000
a. Dependent Variable: Attitude

Model 2: 𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑖 = 𝛽0 + 𝛽1 Female𝑖 + 𝛽2 Color𝑖 + 𝛽3 (Color𝑖 × Female𝑖 )


For which the following SPSS linear regression output was obtained:

ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression ___A__ __B__ __D__ 189 ,000b
Residual 296 __E__ 1
Total 862 299
a. Dependent Variable: Attitude
b. Predictors: (Constant), Female, Color, Female*Color

Coefficientsa
Standardized
Unstandardized Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 2,749 ,100 27,518 ,000
Female 2,397 ,141 ,707 16,967 ,000
Color -1,433 ,173 -,398 -8,283 ,000
Female*Color -,028 ,245 -,006 -,116 ,908
a. Dependent Variable: Attitude

Model 3: 𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑖 = 𝛽0 + 𝛽1 Early Interruption𝑖 + 𝛽2 Middle Interruption𝑖 + 𝛽3 Late Interruption𝑖


For which the following SPSS ANOVA output was obtained:
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 11,083 3 3,694 1,284 ,280
Within Groups 851,368 296 2,876
Total 862,451 299

a) Some part of the ANOVA output for model 2 is missing (now replaced by A, B, C, and D). Fill
in the missing numbers

b) The following pictures show: (1 and 2) the marginal distributions (histograms) resulting from
implementing any of the three models (top) vs. the distribution (histogram) of the dependent
variable (bottom), and (3) a bar plot resulting from implementing one of the three models. 𝑋1
and 𝑋2 denote two dummy variables (these may be Female𝑖 , Color𝑖 , Music𝑖 , etc.) Match each
pictures (1, 2, and 3) with the corresponding model (model 1, model 2 and model 3) based on
their respective SPSS output. Explain your answer. No points will be awarded if you match the
picture correctly, but no motivation is provided.

Picture 1.
Picture 2.

Picture 3
We modify model 2 as follows and obtaining the following output:
log (𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑖 ) = 𝛽0 + 𝛽1 Female𝑖 + 𝛽2 Color𝑖 + 𝛽3 Color𝑖 × Female𝑖

Coefficientsa
Standardized
Unstandardized Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) ,5 ,053 8,816 ,000
Female ,9 ,075 ,619 11,754 ,000
Color ,5 ,092 ,355 5,834 ,000
Female*Color -,1 ,129 -,051 -,750 ,454
a. Dependent Variable: LogY

Consider the revised model 2, with log (𝐴𝑡𝑡𝑖𝑡𝑢𝑑𝑒𝑖 ) as dependent variable:


c) What is the expected attitude (not the log(attitude)) of a male when an advertisement has
bright, flashy colors? Remember: exp(log(X)) = X.
d) Given that the advertisement has bright, flashy colors, what is the expected percentage change
in attitude when the advertisement is shown to females, instead of males?
e) Which variable has the highest (positive) impact on attitude?

Question 5
In this question you’ll have a gentle introduction to dynamic effects. We will consider the evolution of
chart positioning of songs on the Billboard Hot 100. The latest ranking is available here:
https://www.billboard.com/charts/hot-100

The file Top100Billboard2005.csv contains the songs charting in the Billboard 100 during 2005 (the
period considered ranges from June 2004 to June 2006, to account for songs that have been charting
before January 2005 or later than December 2005). The file contains the following variables:
• url: corresponding url where the data was downloaded
• Date: week id of the song
• Week Position: position of the song on the billboard Hot 100
• Song: id of the song
• Performer: id of the performer
• SongID: a unique id to identify the song (the same song may be sung by different performers)
• Instance: this is used to separate breaks on the chart for a given song. For instance, an instance of 6 tells you that
this is the sixth time this song has appeared on the chart
• Peak Position: as of the corresponding week
• Weeks on chart: as of the corresponding weak

a) Select from the dataset the observations concerning the song “Mr. Brightside” by the Killers.
Plot the chart position (i.e. Week position) of the song over time (i.e. Weeks on chart). Interpret
the graph.
b) Next, plot the a graph with the chart position of the song at time t on the y-axis against the chart
position at time t-1 (the previous period) on the x-axis
c) Select from the dataset the observations concerning the song “Inside your heaven” by Carrie
Underwood. Do again a) and b) for this song.
d) Estimate the following model for the song “Mr. Brightside” and interpret the coefficients:
𝑊𝑒𝑒𝑘 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑡 = 𝛽0 + 𝛽1 𝑊𝑒𝑒𝑘 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑡−1 + 𝜀𝑡
Where 𝑊𝑒𝑒𝑘 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑡 is the song position at time t and 𝑊𝑒𝑒𝑘 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑡−1is the song position at
time t-1 (the previous week).
e) Try adding more lags gradually (that is, 𝑊𝑒𝑒𝑘 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑡−2 – 2 weeks ago, 𝑊𝑒𝑒𝑘 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑡−3
– 3 weeks ago, etc. up to t-5). Discuss how the fit of the model changes and interpret the
coefficients.
f) Suppose that we were interested in predicting the chart positioning for multiple songs. That is,
we are interested in predicting chart positioning 𝐶ℎ𝑎𝑟𝑡 𝑃𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑖𝑡 for song i at time t (this is
technically known as a panel). Based on the patterns in question b) and question c), can you
think of a suitable model to do that? No need to implement the model in SPSS, just indicate a
suitable model.

You might also like