A hybrid choice model with a nonlinear utility function and bounded distribution for latent variables: application to purchase intention decisions of electric cars

Research output: Contribution to journalArticleAcademicpeer-review

7 Citations (Scopus)
129 Downloads (Pure)

Abstract

The hybrid choice model (HCM) provides a powerful framework to account for heterogeneity across decision-makers in terms of different underlying latent attitudes. Typically, effects of the latent attitudes have been represented assuming linear utility functions. In contributing to the further elaboration of HCMs, this study suggests an extended HCM framework allowing for nonlinear utility functions of choice alternatives including not only observed but also latent variables. Box–Cox transformations are used to represent the nonlinear utility function. Johnson’s SB distribution is suggested to represent the random term of the latent variables, satisfying the constraint of the Box–Cox transformation. An empirical study using stated choice data about the intention to purchase electric cars is conducted. The empirical results show that the proposed framework can capture nonlinear effects of underlying variables including latent attitudes, thereby enhancing the explanatory power of the choice model.

Original languageEnglish
Pages (from-to) 909-932
Number of pages24
JournalTransportmetrica A: Transport Science
Volume12
Issue number10
DOIs
Publication statusPublished - 21 Jun 2016

Fingerprint

purchase
Railroad cars
decision maker

Keywords

  • bounded distribution
  • Box–Cox transformation
  • electric car
  • Hybrid choice model
  • latent attitudes
  • nonlinear utility function

Cite this

@article{550066039ffa4c678ad098a2f3f1a605,
title = "A hybrid choice model with a nonlinear utility function and bounded distribution for latent variables: application to purchase intention decisions of electric cars",
abstract = "The hybrid choice model (HCM) provides a powerful framework to account for heterogeneity across decision-makers in terms of different underlying latent attitudes. Typically, effects of the latent attitudes have been represented assuming linear utility functions. In contributing to the further elaboration of HCMs, this study suggests an extended HCM framework allowing for nonlinear utility functions of choice alternatives including not only observed but also latent variables. Box–Cox transformations are used to represent the nonlinear utility function. Johnson’s SB distribution is suggested to represent the random term of the latent variables, satisfying the constraint of the Box–Cox transformation. An empirical study using stated choice data about the intention to purchase electric cars is conducted. The empirical results show that the proposed framework can capture nonlinear effects of underlying variables including latent attitudes, thereby enhancing the explanatory power of the choice model.",
keywords = "bounded distribution, Box–Cox transformation, electric car, Hybrid choice model, latent attitudes, nonlinear utility function",
author = "J. Kim and S. Rasouli and H.J.P. Timmermans",
year = "2016",
month = "6",
day = "21",
doi = "10.1080/23249935.2016.1193567",
language = "English",
volume = "12",
pages = "909--932",
journal = "Transportmetrica A: Transport Science",
issn = "2324-9935",
publisher = "Taylor and Francis Ltd.",
number = "10",

}

TY - JOUR

T1 - A hybrid choice model with a nonlinear utility function and bounded distribution for latent variables

T2 - application to purchase intention decisions of electric cars

AU - Kim, J.

AU - Rasouli, S.

AU - Timmermans, H.J.P.

PY - 2016/6/21

Y1 - 2016/6/21

N2 - The hybrid choice model (HCM) provides a powerful framework to account for heterogeneity across decision-makers in terms of different underlying latent attitudes. Typically, effects of the latent attitudes have been represented assuming linear utility functions. In contributing to the further elaboration of HCMs, this study suggests an extended HCM framework allowing for nonlinear utility functions of choice alternatives including not only observed but also latent variables. Box–Cox transformations are used to represent the nonlinear utility function. Johnson’s SB distribution is suggested to represent the random term of the latent variables, satisfying the constraint of the Box–Cox transformation. An empirical study using stated choice data about the intention to purchase electric cars is conducted. The empirical results show that the proposed framework can capture nonlinear effects of underlying variables including latent attitudes, thereby enhancing the explanatory power of the choice model.

AB - The hybrid choice model (HCM) provides a powerful framework to account for heterogeneity across decision-makers in terms of different underlying latent attitudes. Typically, effects of the latent attitudes have been represented assuming linear utility functions. In contributing to the further elaboration of HCMs, this study suggests an extended HCM framework allowing for nonlinear utility functions of choice alternatives including not only observed but also latent variables. Box–Cox transformations are used to represent the nonlinear utility function. Johnson’s SB distribution is suggested to represent the random term of the latent variables, satisfying the constraint of the Box–Cox transformation. An empirical study using stated choice data about the intention to purchase electric cars is conducted. The empirical results show that the proposed framework can capture nonlinear effects of underlying variables including latent attitudes, thereby enhancing the explanatory power of the choice model.

KW - bounded distribution

KW - Box–Cox transformation

KW - electric car

KW - Hybrid choice model

KW - latent attitudes

KW - nonlinear utility function

UR - http://www.scopus.com/inward/record.url?scp=84975263519&partnerID=8YFLogxK

U2 - 10.1080/23249935.2016.1193567

DO - 10.1080/23249935.2016.1193567

M3 - Article

AN - SCOPUS:84975263519

VL - 12

SP - 909

EP - 932

JO - Transportmetrica A: Transport Science

JF - Transportmetrica A: Transport Science

SN - 2324-9935

IS - 10

ER -