عنوان مقاله [English]
The present research used a sophisticated model (mixed logit model) to estimate children's school travel mode use instead of last simpler analysis (more confined models) and other statistical analysis. In the past school travel studies, heterogeneity around the mean of variables in the decision maker's behavior has not been identified. Hence, this study attempted to identify parental taste variation on their children mode choice. The study area was in Rasht, Iran. A questionnaire was designed and 1078 questionnaires were distributed among parents of students from grade one to three in nine elementary schools (including public and private boys and girls schools) in all districts of Rasht' education department and all socio-economic status of the city in January 2014 (Return Rate: 80\%). Based on valid observation (712 individual) from a field survey, results showed that mixed logit (ML) model has better goodness of fit compared with multinomial logit (MNL) model in 99\% confidence interval. In ML model, the effect of home to school distance had a heterogeneity with normal distribution (mean= -0.24 and SD= 0.08) instead of a mean (-0.12) in the MNL model. One percent increase in distance reduces the probability of walking mode by around 0.78 and 0.86 in MNL and ML models, respectively. Furthermore, ML results showed that households, with two or more owned cars, had more sensitivity to walking mode in children's school trip. Some possible parts of the difference in taste variation towards perceived distance of walking from home to school in the walking utility function could be explained by the number of vehicles owned by households. Further, by increasing the walking time from home to school, the probability of choosing walking mode to school will be reduced over school service and household private car. Thus, the existence of high quality school (e.g., well-educated teachers) in all areas of Rasht might not solve the probability of choosing walking mode versus choosing motorized modes.