叙述性偏好法实验的有效样本量模型框架

Framework of Effective Sample Size Model for Stated Preference Experiment

摘要:

叙述性偏好法（SP）是一种通过实验设计研究人们偏好的方法，通常结合离散选择模型估计参数.针对SP实验设计缺乏足以得到准确参数估计的有效样本量的问题，提出一个易于操作、更加全面的SP实验有效样本量估计方法框架，实现通过一个线性模型来估计实验所需的有效样本量.应用该框架开展实证研究，发现要素的数量、要素水平的数量、参数的尺度、样本量以及实验设计策略都对参数的准确性产生显著的影响，得到3种设计策略下的有效样本量估计模型.以此为基础，提出SP实验设计的实践原则.

Abstract:

The stated preference (SP) method is a way to study people’s preferences through experiment design, often implemented with discrete choice models to estimate the parameters. This paper focuses on deriving the effective sample size for SP experiments which guarantees accurate parameter estimations by proposing an easy-to-operate and comprehensive framework. By applying the framework on an empirical study, a linear model is estimated which reveals that the number of factors and levels, scale of parameters, sample size, and the experimental design strategy have significant impacts on the accuracy of parameters. The effective sample size estimation models of three design strategies are obtained, based on which some practical principles for SP experiment design are proposed.