• Home
  • Textbooks
  • Statistics for Marketing and Consumer Research
  • Discrete Choice Models

Statistics for Marketing and Consumer Research

Mario Mazzocchi

Chapter 16

Discrete Choice Models - all with Video Answers

Educators

Chapter Questions

02:17

Problem 1

Open the EFS data-set
a. Estimate a binary logistic regression where the possession of a computer (a1661) is related to income (incanon), age of the household reference person (p396) and household size (a049)
b. Which of the variables is the most important in predicting computer possession?
c. How many cases are correctly predicted by the model?
d. Estimate the model with ordinary linear regression and compare the results
e. Estimate the model with a probit regression and compare the results
f. Replicate the results using the generalized linear model
g. Include NS-SEC 8 Class of Household Reference Person (a094) as a categorical predictor and the interaction between age and income. Do results improve?

Victor Salazar
Victor Salazar
Numerade Educator
02:17

Problem 2

Open the EFS data-set
a. Recode the tenure-type variable (a121) into three categories, owned (7), owned with mortgage (5), rented (1 to 4 ) and treat other values as missing
b. Explain the variable with a multinomial logistic regression using the following predictors (covariates):
i. Income (incanon)
ii. Age of the household reference person (p396)
iii. Household size (a049)
iv. Recreation expenditure (p609)
c. Use the stepwise method to maintain the relevant variables
d. Is the model satisfactory in terms of correct prediction rate?

Victor Salazar
Victor Salazar
Numerade Educator
03:12

Problem 3

Open the Trust data-set
a. Estimate an ordered probit model to explain how the likelihood to purchase chicken (q7) is explained by the determinants of the Theory of Planned Behavior (see box 15.3) - HINT: include them in the model as covariates:
i. Attitudes toward chicken ( $\mathrm{q} 9$ )
ii. Subjective norm (SN)
iii. Perceived behavioral control (PBC)
b. Which variable is the most relevant in explaining the likelihood to purchase?
c. What is the rate of correct prediction?
d. Add average price (PRICE) and income (INC) and check the model performance.

Shu Naito
Shu Naito
Numerade Educator