Citation

Isihara P, Shi C, Ward J, O'Malley L, Laney S, Diedrichs D, Flores G. Journal of Choice Modelling 2020; 35: e100204.

Copyright

(Copyright © 2020, Elsevier Publishing)

DOI

10.1016/j.jocm.2020.100204

PMID

unavailable

Abstract

This paper proposes the use of Most Typical (MT) and Most Ideal (MI) levels when an adaptive choice-based conjoint (ACBC) survey can only obtain a small sample size n from a small population size N. This situation arises when expert decision makers are surveyed from among important small populations such as executives of large companies or political leaders, for which the expert decision maker assumption is reasonable. The paper compares respondents' MT levels obtained using the Build Your Own (BYO) question with MI levels obtained using part-worth utilities. The MI levels are validated using the Potentially All Pairwise RanKings of all possible Alternatives (PAPRIKA) method. It then explores differences in MT/MI levels for two related populations using an application concerning disaster relief. For effective disaster relief coordination, humanitarian organizations must understand each other's response decisions. An ACBC survey on the "Go/No-Go" decision by 49 faith-based (FBOs) and 12 non faith-based (NFBOs) disaster relief organizations considered four attributes: Funding, Disaster Response Type, Need Assessment, and Community Access. There was disparity between MT/MI Funding levels: 18 of 19 respondents reported MT levels of 50% or less, but 12 of 19 estimated to have MI levels of at least 75%. Greatest similarity between FBOs and NFBOs was observed for MI Need Assessment. Greatest disagreement of MI levels determined by part-worths and PAPRIKA was for Need Assessment and Disaster Response Type. To handle zero counts in the sample frequency distributions, we include a mathematical appendix explaining our use of a Bayesian rather than maximum likelihood estimation of MT/MI population frequency distributions.

Language: en

Keywords

ACBC survey of disaster relief “Go/No-Go” decision; Bayesian estimation of multivariate hypergeometric distribution zero counts; Most typical (MT) / most ideal (MI) attribute levels; Part-worth utilities; Potentially all pairwise RanKings of all possible alternatives (PAPRIKA) method