Module 4 (Lecture 4, Articles)

Question 1

What are some limitations of scanner data?

Answer 1

1. Small frame, small shops may not be considered.
2. Cannot make causal statements right away.
3. Don’t know behaviours and psychographics.
4. Don’t know exact set of choices, prices, and the time of decision.

Question 2

What can brands perception data tell us and what are some limitations?

Answer 2

Several market research companies track the perception of firms and brands. This includes variables such as attitudes towards towards the firm, customer satisfaction, reputations. This indicates how strong a brand is in the hearts and minds of consumers.

Limitations are that it is not reflecting actual purchase behaviour. Also response bias, sampling bias.

Question 3

What is response bias?

Answer 3

When participants give inaccurate or dishonest answers to self-report questions.

Question 4

What is sampling bias?

Answer 4

When the selected sample is not representative of the target population.

Question 5

What are three essential steps in statistical data editing?

Answer 5

1. Error localisation: determine which values are wrong.
2. Correction: correct missing and wrong data in best possible way.
3. Consistency: make sure everything is consistent without conflicts.

Question 6

Interviewer error (why data editing is important)

Answer 6

Interviewers may not be giving the respondents the correct instructions.

Question 7

Omissions (why data editing is important)

Answer 7

Respondents often fail to answer a single question or a section of the questionnaire.

Question 8

Ambiguity (why data editing is important)

Answer 8

A response might not be readable or it might be unclear.

Question 9

Lack of cooperation (why data editing is important)

Answer 9

In a long questionnaire, a respondent might rebel and check the same response.

Question 10

Ineligible respondent (why data editing is important)

Answer 10

An inappropriate respondent may be included in sample, e.g. underage.

Question 11

What is data coding?

Answer 11

You specify how the information should be categorised to facilitate the analysis.

Question 12

What is data matching?

Answer 12

Combining data from multiple sources that refer to the same entities. So you identify, match, and merge records.

Question 13

What is data imputation?

Answer 13

The process of estimating missing data and filling in these values.

Question 14

What is data adjusting?

Answer 14

Process to enhance the quality of the data for the data analysis (e.g. weighting, variable respectification, scale transformation).

Question 15

What is weighting? (In data adjusting)

Answer 15

Adjusting the influence of certain observations so that the sample better reflects the population.

Question 16

What is variable respectification? (In data adjusting)

Answer 16

The process of restructuring existing variables. Sometimes variables are too detailed, or you have too many similar categories.

Question 17

What is scale transformation? (In data adjusting)

Answer 17

Process of adjusting the scale to make sure it is comparable with other scales. E.g. Some people consistently use the low end, and others the high-end, even if they feel the same.

Question 18

Entity extraction (a task to prepare text data)

Answer 18

Which words people write about?

Question 19

Topic modelling (a task to prepare text data)

Answer 19

What topics people write about?

Question 20

Sentiment analysis (a task to prepare text data)

Answer 20

How positive/negative is the text?

Question 21

What are the two different ways we can use text data?

Answer 21

1. Language reflects = language is used to understand people, it reflects what they think, feel, and do.
2. Language affects = language can influence outcomes, it changes how people think, feel, act.

Question 22

Difference between mode, median, and mean?

Answer 22

Mode = most frequent value.

Median = value that lies in the middle of a frequency distribution.

Mean = average value (sum/number of numbers).

Question 23

Which descriptive statistics can be used with nominal, ordinal, and interval/ratio data?

Answer 23

Nominal = mode.
Ordinal = median, mode.
Interval/ ratio = mean, median, mode.

Question 24

Imagine a correlation of 0.55 between ad expenditures and sales.

Does this mean more ad expenditures result in higher sales?

Answer 24

No, correlation only shows if the variables move together, not why.

There may be:
1. Spurious correlation = a third factor/confounding variable causes both.
2. Reverse causality = ad expenditures are determined as a % of sales, so sales determine ad expenditures, not the other way!

Question 25

What are the three conditions for causal relationships?

Answer 25

1. Strong association (e.g. correlation). 2. Time order: cause happens before effect (this rules out reverse correlation!). 3. No other explanation possible (no confounding/third variables). Best tested in experiments.

Question 26

Experimental group

Answer 26

Test subjects who were exposed to the experimental stimulus.

Question 27

Control group (in an experiment)

Answer 27

Test subjects who are not exposed to the experimental stimulus.

Question 28

Randomising (in an experiment)

Answer 28

Random assignment of test subjects to experimental/control groups.

Question 29

Matching (in an experiment)

Answer 29

Test subjects in experimental and control groups share specific criteria. This to ensure that differences in outcomes between the groups can be attributed to the presence of the stimulus, not other factors.

Question 30

Stimulus (in an experiment)

Answer 30

Change in a variable that should trigger a behavioral reaction in people.

Question 31

What are the advantages of lab experiments (2) and field experiments (2)?

Answer 31

Lab experiments: 1. Higher internal validity, because stimuli can be more effectively manipulated and control factors better controlled. 2. Lower costs. Field experiments: 1. Higher external validity, because test subjects are acting under real conditions. 2. Easier to predict and generalize the effect.

Question 32

What are the disadvantages of lab experiments (2) and field experiments (4)?

Answer 32

Lab experiments: 1. Test subjects do not react exactly as in a natural environment; making generalizations and predictions of effect hard. 2. Lower external validity. Field experiments: 1. Costly 2. Activities visible to competitors. 3. Less manipulation freedom. 4. More difficult to control external factors.

Question 33

Internal vs. external validity

Answer 33

Internal validity = how well a study establishes a causal relationship between variables by minimising confounding variables and bias. External validity = the extent to which the study results can be generalized to other populations and settings.

Question 34

What are 3 benefits of using causal models in marketing research?

Answer 34

1. They clarify which variables cause what. 2. They make definitions and measurements more precise. 3. They help represent complex theories with many variables and relationships.

Question 35

4 challenges in causal models marketing research?

Answer 35

1. Replicability issues: poor documentation of causal models, two-thirds lacked sufficient data for replication. 2. Small sample sizes: half of the models used less than 200 data points, leading to unstable results. 3. Measurement issues: many studies used too few indicators per construct (e.g. no multi-item), making it difficult to assess reliability and validity. 4. Two stage estimation: most researchers used a two stage approach (instead of one stage), which separates measurement and theory.

Question 36

What is signaling theory?

Answer 36

Signals can be used to communicate information that is not directly observable. E.g. giveaways may signal to consumers that companies are investing in the consumer-brand relationship. However, also negative consumer reactions are possible.

Question 37

Prospect theory regarding giveaways

Answer 37

Prospect theory suggests that consumers perceive losses more strongly than gains, meaning that a low-quality giveaway has a greater negative impact on consumer attitudes than a high-quality giveaway has a positive one.