1993
DOI: 10.1287/mksc.12.3.213
|View full text |Cite
|
Sign up to set email alerts
|

Modeling Multiple Sources of Heterogeneity in Multinomial Logit Models: Methodological and Managerial Issues

Abstract: We offer a framework to specify and estimate various sources of heterogeneity in multinomial logit brand choice models. We let each brand-specific intercept and each parameter of the explanatory variable vector vary randomly across households. In addition, we distinguish loyal households from the rest. Our results suggest that incorporating multiple sources of heterogeneity improves the model fit and suggests higher impact of marketing mix elements on brand choice. We highlight the importance of incorporating … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
34
0

Year Published

1998
1998
2013
2013

Publication Types

Select...
9
1

Relationship

0
10

Authors

Journals

citations
Cited by 184 publications
(34 citation statements)
references
References 35 publications
0
34
0
Order By: Relevance
“…Specifically, if menu m = q 1 q 2 q n is offered with prices p = p 1 p 2 p n , the probabilities of buying variant j = 1 n, or the outside good are Gönül and Srinivasan (1993), who give a general discussion of Bayesian logit models and random brand effects, describe the randomness of b 0 and b 1 as capturing "variations on the intrinsic brand utility across households" (p. 216). We further assume that E e − < + , which is satisfied by the normal distribution and any distribution with bounded support.…”
Section: Assumption 5 Demand For Offered Variants and The Outside Gomentioning
confidence: 99%
“…Specifically, if menu m = q 1 q 2 q n is offered with prices p = p 1 p 2 p n , the probabilities of buying variant j = 1 n, or the outside good are Gönül and Srinivasan (1993), who give a general discussion of Bayesian logit models and random brand effects, describe the randomness of b 0 and b 1 as capturing "variations on the intrinsic brand utility across households" (p. 216). We further assume that E e − < + , which is satisfied by the normal distribution and any distribution with bounded support.…”
Section: Assumption 5 Demand For Offered Variants and The Outside Gomentioning
confidence: 99%
“…a careless driver has a higher probability of being involved in a car accident). Kamakura and Russell [10] show how to segment a market into loyal and nonloyal customers, where the loyal customers are less price sensitive and keep buying the same brand (see the works of [6] and [20] who show that even a very short purchase history data can have a huge impact, and that of [16] which shows that consuming one product from a particular company increases the probability of consuming another product from the same company).…”
Section: Related Workmentioning
confidence: 99%
“…The mixed multinomial logit (MNL) model specified above and its variant have been well studied and widely applied in economics and marketing (e.g., Guadagni and Little 1983, Kamakura and Russell 1989, Chintagunta et al 1991, Gönül and Srinivasan 1993. 7 The Bayesian approach and the MCMC algorithm for the mixed MNL estimation are well established Allenby 1993, Allenby andLenk 1994).…”
Section: Ketchup Data Setmentioning
confidence: 99%