Module 6 (Lecture 6, Articles, Tutorial 5) Flashcards
(23 cards)
What is a sales response model?
Sales response models try to model a sales response as a function of business activities.
What are the advantages and disadvantages of internal data?
Advantages = usually readily available, accurate, and quick.
Disadvantages = there is no data on data competitors, no info on time of decision or psychographics.
E.g. you might want to know how competitors react to business strategies you conduct as a company.
What are the advantages and disadvantages of annual reports?
Advantages = available for all companies listed on a stock market.
Disadvantages = data only available on yearly or quarterly basis. Non-traded companies not considered. No info about psychographics or time of decision.
What are the 5 types of sales response models? And give the exact names of models.
- Constant marginal returns = linear model.
- Decreasing marginal returns = multiplicative model, semi-logarithmic model.
- Saturation volume = modified exponential model.
- S-shaped = log-reciprocal model, logistic model.
- Market share models = multiplicative interaction model, multinomial logit model.
What is elasticity?
Elasticities express the relative change in one variable (e.g. sales volume) caused by a relative change in another variable (e.g. advertising).
What could you do if x is close to zero in a semi-logarithmic model?
It’s very common to add a small number, e.g. 1 to the x variable.
What is an advantage of the multiplicative model with regard to beta?
We can see beta = elasticity right away in the formula because it’s an exponent.
So you don’t really need to calculate elasticity in this specific model.
What is the threshold effect in S-shaped models (log-reciprocal model, logistic model)?
In case of e.g. advertising spending, spending is not effective until they exceed a minimum level, then they go up heavily, then decreasing marginal returns. This is the S shape.
What is a logistic model?
A logistic model is used when the DV is binary (0/1 or yes/no).
It predicts the probability that a specific event will happen.
This value is always between 0 and 1.
What are the 5 implications of a logistic model?
- Probabilities must lie between 0 and 1.
- Non-linear regression line - often S-shaped.
- Error term needs to follow a logistic distribution.
- You cannot use the least squares method like in a linear regression -> use a maximum likelihood estimation.
- R^2 from linear regressions also doesn’t work here.
What is a carry over effect?
Customers, retailers and competitors might need time to react to market activities (delayed response effects).
Especially the impact of advertising is considered to be a dynamic process.
(Note: not a carry over effect, but a dynamic effect is when people react to marketing activities in advance).
What does the linear sales response model assume?
Constant marginal returns
What are the advantages and disadvantages of linear sales response model?
Advantages:
- Can be estimated easily by regression analysis.
- Simple and easy to understand.
Disadvantages:
- Constant returns to scale not realistic.
- Assumption that sales can be infinitely increased not realistic.
What do the multiplicative model and semi-logarithmic model assume?
Decreasing marginal returns.
What are the advantages and disadvantages of the multiplicative model and semi-logarithmic model?
Advantages:
- Diminishing marginal returns more realistic.
- Easy linearization for the semi-logarithmic model with the help of a logarithm.
- In the multiplicative model, the elasticity can be derived directly from the exponent.
Disadvantages:
- No saturation level.
- For semi-logarithmic model: if x is close to 0, sales would go to minus infinity, which is impossible.
What are the advantages of a modified exponential model?
- Saturation level corresponds to realistic consumer behavior.
No disadvantages!
What is the advantage and disadvantages of the log-reciprocal model and logistic model?
Advantage:
- Accounts for the phenomenon that advertising needs to be raised above a certain level to have an impact.
Disadvantage:
- For logistic model: DV is only predicted in the form of a probability.
What are the 5 differences between a linear regression and a logistic model?
- Linear regression for a continuous DV and a logistic model for a discrete DV.
- In a linear regression there needs to be a linear relationship between DV and IV. For logistic model not the case.
- For a logistic model, DV is between 0-1. For linear regression no restriction.
- In a linear regression, normal distribution of DV is assumed. In a logistic model, binomial distribution of DV is assumed.
- Linear regression is a straight line. Logistic model is S-shaped.
When is something perfectly inelastic?
Elasticity = 0.
Outcome variable does not change at all.
E.g. a 10% price increase leads to 0% change in sales.
When is something inelastic?
0 < elasticity < 1.
Outcome changes but less than proportionally compared to input.
E.g. a 10% increase in price causes only a 5% drop in sales (e = -0.5).
When is something unit elastic?
Elasticity = 1.
E.g. a 10% price drop leads to a 10% increase in sales.
When is something elastic?
Elasticity > 1.
Outcome changes more than proportionally in response to the input.
E.g. a 10% increase in advertising, results in a 15% increase in sales (e =1.5).
If you see a sales response model with ln, which specific model is this?
Semi-logarithmic model.
