Raimund K. Herz

The Use of the Bicycle

Homepage Manual Contents Navigation

APPROACHES FOR ANALYSING AND MODELLING THE TRAVEL BEHAVIOUR OF CYCLISTS

Of course, there is no subspecies "cyclist" among human beings. Cyclists are ordinary individuals pursuing their activities more or less voluntarily under a set of aspirations and constraints. Under specific circumstances, the resulting need to move from one place of activity to another is realized with the help of a bicycle. Otherwise other modes of transport are used. Consequently we have to look at individuals, their activity programmes and the constraints they have particularly with respect to the choice of different modes of transport. It is only at this micro-level that the real reasons for using or not using the bicycle appear. At an aggregate level of situational, personal or spatial categories, parts of the explanatory power are averaged out and may even completely vanish. There must be some aggregation. However, the preferences are: as little aggregation over the explanatory variables as possible, and: the later the aggregation within the analytical and modelling process the better.

The two approaches with least diluting effects of aggregation are random utility models and Brög's analytical concept of situational constraints. They are well documented elsewhere.^7-9

At first sight, random utility models seem to be quite appropriate for analysing the modal choice of the bicycle. The alternatives are clearly defined and not too many: walking, using public passenger transport. driving or riding a private car or motorcycle. There is no lack of attributes neither of these alternative modes nor of the individual, for furnishing the utility function. Apart from the assumptions about the distribution of the error term, problems arise from the difficulty of gathering information on all of these attributes of the alternative modes of which an individual is probably unaware. The most crucial fact however is that cyclists are mainly, and for different reasons, captive riders. The situation of free choice exists only for a small minority. On the other hand. the choice set might as well be expanded: human behaviour is flexible enough. at least with respect to non-duty activities. to reschedule daily programmes, to redistribute role functions and to change activity places in order to use a more convenient mode of transport. Anyway. in the utility function for a restricted set of alternatives the most relevant attributes. especially those characterizing the individual and his personal situation, are not substitutable. Market segmentation and hierarchical modelling seem to lead only half-way out of this dilemma.

On the other hand, Brög's situational grouping approach relies heavily on the art of interactive measurement¹⁰ for all those soft ingredients defining and restricting the field of choice. The analyst must decide from the answers of the respondents whether he or she is still free to choose the bicycle for a specific trip or not. Sometimes this analytical decision is an easy one, sometimes the distinction may be quite arbitrary and difficult to document. Grouping according to policy relevant constraints, which may be removed by respective measures. is certainly a major advantage of this concept compared with traditional socio- economic segmentation. In this approach. the complex situation of the individual is not cut into pieces but preserved for analysis. However, by embracing all kinds of personal and locational characteristics, the results do not lend themselves directly for social and physical planning in the traditional sense. It may be difficult to identify target groups within these situational groups, which are homogeneous only with respect to the specific context of the decision. not with respect to specific social characteristics or locations.

A third and more conventional approach is category analysis. The problems with this approach are also manifold. First, it needs a relatively large sample in order to fulfill the minimum of entries required for all categories. Second. theoretical or hypothetical and empirical knowledge is required for establishing those categories which are not only most discriminant with respect to bicycle use but also most appropriate for planning and policy making. Third. there is a dilemma between clear-cut categories and the variation of the behavioural variable: a complex structure of the category will be necessary for yielding homogeneous behaviour, as in the case of Brög's situational categories. Fourth and last, categories of persons do not really exist. They are artificial creations with an average behaviour. The behavioural variation within is usually much greater than that between these categories.¹¹

Nevertheless, and in spite of the fact that the explanatory power is considerably smaller on this level of aggregation, we start with this sort of analysis, setting the scene for subsequent analyses of individual bicycle choice behaviour.