考虑心理潜变量的出行方式选择行为模型

Incorporating Psychological Latent Variables into Travel Mode Choice Model

为系统考虑心理因素对出行方式选择行为的影响,以计划行为理论为框架整合了对通勤方式选择行为具有影响的各种心理因素,分别对摩托车、公交车和小汽车构建了3种多指标多原因模型,估计模型参数后预测心理影响潜变量的拟合值,作为解释自变量加入到传统的条件Logit模型中形成了混合选择模型。采用实证数据分别以公交车和小汽车作为效用基础项估计并比较了传统离散选择模型和混合选择模型中各参数影响的异同,最后利用月收入的变动检验了2个模型的预测能力。结果表明:不是所有的心理影响潜变量都对通勤方式选择行为有显著性影响,混合选择模型比传统不带潜变量的选择模型对实证数据具有更高的拟合度,其预测值也更为稳健。

To systematically research the impacts of psychological factors on travel mode choice behavior,the paper constructed three multiple indicators and multiple causes（MIMIC）models for motorcycle,bus and car,according to the expanded theory of planned behavior（TPB）with variety of psychological factors included with respect to commuter mode choice behavior.The fitted values of latent variables from estimation of MIMIC were introduced into the conditional logit model（CLM）as explanatory independent variables,so as to establish a hybrid choice model（HCM）.The empirical data were collected and used to compare the different influences of parameters between the traditional discrete choice model（DCM）and HCM.The predication capabilities of the two models were tested by using the change of monthly income.The results show that not all of the psychological latent variables have significant effect on commute mode choice behavior.The outcome of estimation of HCM has a higher degree of fitting and much more robust prediction than the traditional DCM without latent variables.