Analysis of travel behavior of households sensitive to environmental justice

For at least the last two decades, environmental and social impacts, and recently, the distributional impacts of decisions regarding transportation infrastructure are better understood by governmental agencies and reflected for funding eligibility. In principle, each transportation project undertaken has to improve existing conditions and should not have disproportionate adverse impacts on minority and low-income populations. Such analysis is a part of an environmental justice exercise. The need for understanding the differences, particularly in terms of travel behavior and those factors that affect travel behavior, between the disadvantaged populations and the rest is imperative. With this objective in mind, a structural equation model has been formulated to identify the differences in relationships that are influential to the household travel behavior of two population groups defined by racial composition and poverty levels of the area of residence. This model represents the following five factors: household structure, urban form, accessibility, transit availability and auto dependence, and interrelationships. The Chicago Area Transportation Study 1990 Household Travel Survey data and regional land use data have been utilized for the model development. The measurement model indicated the presence of substantial differences in factor composition across groups, particularly for urban form and auto dependence, and rejected the measurement equivalence hypotheses. The structural model investigated the effects of household structure, urban form, accessibility, and transit availability on the auto dependence factor. Model estimates indicated that transit availability is the most influential factor on auto dependence while the effect of urban form, accessibility, and mobility needs as defined by household structure did not have significant effect. The effect of transit availability for the disadvantaged group is found to be about twice that for the remaining population. The overall model performance was reasonably satisfactory which indicated the possibilities for improving both the measurement and structural model specifications.