Proyecto de Reparcelación. El marco normativo actual

The above review of literature has underlined some significant gaps on which this study attempts to shed a light on. We categorize them into two specific classes: the quality of data, and analytical methods. Data availability often pre-determines the study method so the two are highly related. Below we are going through them in turn.

2.4.1. Quality of data

It would seem that in two aspects the current studies still have under-tapped potential which can be associated with the prevailing data difficulties. First, most studies reveal insights into the influences on distance travelled, but so far only very few do so on trip frequency and travel time; this limits the understanding of influences on travel accessibility and leaves an apparent gap on mobility. Secondly, few existing studies except the census-based longitudinal work could easily provide regular updates going forward without major data efforts. This is foremost a data issue, as few researchers would disregard particular explanatory variables or travel outcomes if suitable data is available.

This issue is well outlined in Preston and Rajé (2007) where they discuss inequality in transport. They show that it is necessary to understand both accessibility and mobility patterns in order to design effective policy responses. The problems of the disadvantaged cannot be analysed in isolation from the rest of society, not least because we need to have a clear understanding of how big the disadvantage gaps are among them. For rich countries where the majority of travel is suburb to suburb, some enjoy fast and smooth car or rail journeys whilst others rely on infrequent, expensive and poorly connected public transport. Such differences could arise from a wide range of causes, such as demographic-socioeconomic circumstances, built form, gender, life-cycles, lifestyles, ownership/access to car, social and environmental attitudes, etc. Furthermore, the circumstances and attitudes could evolve rapidly, given the momentous changes in labour market and wider society.

In this context, it is a little surprising that the potential of UK National Travel Survey (NTS) has not been more fully investigated for this purpose. The NTS has been collecting an extensive household sample dataset since 1965, and since 1988 the survey has been carried out every year. The survey is conducted as home interviews of all household members, recording a detailed one-week travel diary together with carefully selected personal, household and

35

circumstantial variables that are thought to influence travel behaviour. The data is weighted to provide annual updates on all main purposes of domestic travel in terms of travel distances, times and frequency. The list of the variables is arguably the most comprehensive among nation-wide travel surveys.

Hitherto, there are only very limited attempts to relate travel patterns to the extensive range of the NTS variables (Stead and Marshall, 2001, Stead, 2001, Dargay and Hanly, 2004, Jahanshahi et al., 2009, Jahanshahi et al., 2013, Susilo, 2015), and none except the last one have made use of the improved time series of survey results since 2002. Methodological limitations tend to be the main reason that has held back a fuller exploitation of the comprehensive list of NTS variables; the personal, households and circumstantial variables are highly intercorrelated because of self-selection, spatial sorting and other endogeneities. In addition, there may be interactions among trip purposes and travel outcomes (e.g. long distance commuters might travel less frequently and forgo some other trips). Unlocking insights in the data would require robust models that can cope with such complexity.

2.4.2. Conceptual and methodological gaps

There is ample room for improvement in conceptual design and analytical methods regarding the influence of built form influence on travel. First, existing studies collectively suggest that significant endogenous interactions exist among the influencing factors like travellers’

socioeconomic and demographic profiles, residents’ self-selection and spatial sorting, built form and to some extent car ownership; however, few if any studies have examined this whole range of influences in one model.

Secondly, there are some potentially important interactions that have been left under- or un-investigated, such as among different trip purposes or different travel outcomes. For instance, would longer commuting be offset by shorter or fewer shopping journeys, or less frequent travel imply longer distances or durations?

Thirdly, recent studies have benefited from improvements in statistical methods to accommodate heterogeneity among individuals or choices, and characterized the latent states behind individuals’ decisions; however, we are not aware of any study which aims to accommodate these techniques within the SEM for understanding the built form influences.

Categorizing geographical locations through latent class analysis in combination with SEM can

36

better quantify the built form effect to inform built form and transport policies and models. In addition, the incorporation of random intercept models could make it feasible for simultaneously controlling potential endogeneities through SEM and measuring the macro level variations when individuals are nested within more aggregate units. This is the reason for its wide use in other disciplines such as education and health (Cho et al., 2015, Dunn et al., 2015, Marsh et al., 2009, Marsh et al., 2015) where individuals are nested within aggregate units (such as schools in educational studies or built form clusters in our context). It would therefore appear of both theoretical and policy interest to incorporate LCA and random intercept models in SEM for examining the more complex and controversial aspects of influences on travel behaviour.

2.4.3. Summary

The above literature review illustrates the requirement for systematic analyses of comprehensive range of socioeconomic and built form characteristic on all aspects of travel (i.e. accessibility, mobility and frequency of travel) in order to fully comprehend the influences on travel behaviour and measure the extent of built form influences. Not only does this require a high quality travel survey dataset, but also an appropriate methodological approach for modelling highly inter-correlated parameters and accounting for potential endogeneities. This study aims to contribute by employing extended Structural Equation Modelling (SEM) framework to analyse a dataset from the UK National Travel Survey, which is arguably one of the most comprehensive ones, if not the most, in the world. Chapter 3 below provides an extensive explanation of the analytical methods used for this study.

37

3. Method of Study