USING STRUCTURAL EQUATIONS MODELLING TO UNRAVEL THEINFLUENCE OF LAND USE PATTERNS ON TRAVEL BEHAVIOUR OFURBAN ADULT WORKERS OF PUGET SOUND REGION

摘要

Since the 1990s a great number of studies attempted to unravel the ways inrnwhich land use patterns and urban configuration influence travel behaviour.rnImportant methodological advances were made, including but not limited tornthe creation of integrated land-use transportation models, introduction of utilityrnbased models, development of activity based frameworks, andrnmicrosimulation as well. Key to developing good forecasting models isrnunderstanding causalities among the variables used to perform the forecasts.rnDiscovering causalities of this type require statistical tools that can analyzernmany variables simultaneously.rnThis paper addresses the relationship between travel behaviour and land usernpatterns using a Structural Equations Modelling (SEM) framework, which isrnthe best tool for simultaneous equations of dependent and explanatoryrnvariables. SEM is a multi equation technique which is particularly suited forrnthe study of complex relations, since it allows modelling the effects of land usernpatterns on travel behaviour while controlling for self selection bias and effectsrnbetween endogenous variables. In this way we can identify if persons thatrnchoose to live in some places do that because of the type of travel desired orrnif they choose to live in these places and then decide how to travel.rnThe proposed model structure in this paper is by design heavily influenced byrnthe model developed for Lisbon (Abreu e Silva et al, 2006) to allowrncomparisons. In that paper the existence of significant effects of land usernpatterns in travel behaviour was found. The travel behaviour variablesrnincluded in the model are multidimensional and comprehend both short termrnand long term mobility decisions. Regarding long term decisions the modelrnincludes home location characteristics, car ownership levels and transit passrnownership. The shorter term decisions are mobility by mode (car, transit andrnsoft modes), measured in terms of total number of trips. The model alsornincludes a trip scheduling variable, which is the total time spent between thernfirst and last trips.rnThe modelled land use variables measure the levels of urban intensity andrndensity, diversity, both in terms of types of uses and the mix between jobs andrninhabitants and the public transport supply levels,. The land use patterns arerndescribed both at the residence and employment zones of each individualrnincluded in the model. This plethora of variables is reduced to a morernmanageable number of variables using a factor analysis technique.rnIn order to explicitly account for self selection bias the land use variables arernexplicitly modelled as functions of socioeconomic attributes of individuals andrntheir households.