Week 3

Question 1

What is Market Response Model + purpose

Answer 1

• How do changes in MKT activities affect sales and consumer behaviour

It helps predict how customers will respond to different marketing strategies.

Purpose:
*Help managers understand customer responses to marketing activities.
*Provide a macro view of market

Question 2

Applications of Market Response Model (usages) 3

Answer 2

1. Forecast future demand
2. Evaluate past marketing
3. Allocate budgets and resources effectively

Question 3

Regression (purpose + goal + two types + formula)

Answer 3

Purpose: analyse relationships between variables

Goal: make predictions

Two Types:

1. Simple Regression
   - between two variables
2. Multiple Regression
   - between multiple variables

Formula:
y = a + b ⋅ x (+e)
DV = intercept (constant) + slope (B) (change in y for a unit change in x) . IV

Question 4

What is slope? What is intercept?

Answer 4

Slope = indicates the strenght and direction of the relationship between variables

Intercept = expected value of Y (DV) when X is 0 (no IV)

Question 5

How does Simple Regression look graphically

Answer 5

Straight line that defines the relationship between the IV and DV

Question 6

Residual Error

Answer 6

the difference between the actual and predicted values of DV

Question 7

Ordinary Least Squares (OLS)

Answer 7

Draw a line that is as close as possible to all the points on the graph, making sure the overall error is small

Question 8

T-statistic and p-value (def. + thresholds)

Answer 8

T-stat:
whether a result is meaningful or just due to random variation
* t-statistic > 2

p-value:
how likely did the results happen by chance
*p-value 0.05 (likely a real effect)

Question 9

R square (def. + threshold)

Answer 9

shows how well the model fits the data

*closer to 1 the better the fit
*0.70 acceptable R square

Question 10

Adjusted R2

Answer 10

Adjusts R² for the number of IVs and sample size (to increase accuracy)

How much of the change in DV do the IVS account for

(60% Adj.R2 = IV (Ad) is responsible for 60% of the change in sales)

Question 11

Regression Assumptions

Answer 11

Key Assumptions

*Normality: Residuals should be uncorrelated, and normally distributed
*Linearity: Residuals should show a linear relationship
*Homoscedasticity: residuals should constantly show similar patterns

random, constant, linear pattern

Question 12

Problems with Residual Analysis

Answer 12

1. Heteroskedasticity
   - the residuals appear as a FUNNEL-SHAPED pattern (sign of systematic error-continous)
2. Non-linearity
   - when the relationship is not linear (U shaped)

Question 13

Purpose of Multiple Regression + formula

Answer 13

The aim is to predict a DV based on one or more IVs

• Predict
• Explain (the relationship between var)

It extends the simple regression model

Y = a + b1 . x1 + b2 . x2 …. + bn . xn + e

Question 14

Multiple Regression Types

Answer 14

1. Simultaneous Regression: All IVs are entered together
2. Stepwise Regression: when a lot of IVs
   Variables are added/removed in steps

a. Backward Elimination
- Start with all IVs, remove the least significant until all remaining are significant.

b. Forward Selection
- Start with the most correlated IVs, add others one by one if they are significant.

Question 15

Regression Diagnostics

Answer 15

1. SIGNIFICANCE: p-value, t-statistics
2. LOGICAL EQUATION: no unexpected signs (+/-)
3. How well does the MODEL FIT
   adjusted R2, actual vs. predicted (lines)
4. Multicollinearity (VIF, Tolerance)

Question 16

Multicollinearity + how to detect (+thresholds)

Answer 16

If two or more IVs are highly correlated (similar), it can cause unreliability and errors.

VIF >10
-Tells how much multicollinearity is affecting the regression.

Tolerance <0.10
- how unique are IVs and not overlapping with other IVs in the model.

Question 17

Magnitude (expB) + Standardized and Unstandardized Coefficients def

Answer 17

Magnitude = strength

how strong is the relationship between IV and AV

* Standardized Coefficients: help compare different variables within the same model.

* Unstandardized Coefficients: compare the same variable across different models.

Question 18

Data Transfromation

Answer 18

Allows the use of categorical variables in a regression

Code these as Dummy Variables

Question 19

Mediation

Answer 19

Explain how one variable affects another.

Example:
(education (x) - job skills (mediator ) - high salary

Education itself does not mean high salary, but it leads to job skill which results in higher income

Question 20

Moderation

Answer 20

Affects the strength or direction of the relationship between two other variables.

Stress (x) - low social support (moderator) - anxiety (y)

Stress itself can lead to anxiety, but adding a moderator (low social support) strenghtens this relationship, but it occurs without it

Question 21

Types of Mediations (5)

Answer 21

1. Complementary Mediation (partial mediation)
   - Indirect effect and direct effect both exist and point in the same direction (+/-)
2. Competitive Mediation
   - Indirect effect and direct effect both exist and point in opposite directions (+/-)
3. Indirect-Only Mediation (full mediation)
   - Only indirect effect exists, direct effect does not
4. Direct-Only Mediation
   - Only direct effect exists, indirect effect does not
5. No-effect Non-Mediation
   - the mediator has no significant impact on the outcome, the direct relationship between X and Y remains unchanged.

Question 22

Spotlight Analysis

Answer 22

How do diff. Levels of moderator change the effect of IV on DV