This report uses data from the Greater London Transportation Survey (GLTS) to model pedal cycle use in Greater London. The average cycling trip rate per person for a given category of residents of a given area is shown to be the product of two practically independent parameters: participation (proportion of people cycling) and activity rate (number of trips per cyclist). Analysis of the data reveals that the participation parameter is the more important in determining differences between trip rates. Approximate statistical independence between those two parameters is also observed when the population is subdivided into 10 social groups. Cycling varies considerably, with the highest levels for children and students and the lowest for housewives and retired people. Car ownership is also associated with significant differences in cycling within particular social groups but its net effect on the cycling trip rate for all residents of a given area is small. These differences in social characteristics between the populations of different parts of the GLTS area account for only a small part of the spatial variations in cycling trip rates. Nevertheless zonal modelling using the fine mesh of the GLTS traffic zones reveals systematic sectoral and radial patterns in the proportion of the population cycling (though not in the activity rates). Within the 10 social groupings the 5 groups for which this zonal model shows the most systematic variation contribute 75 per cent of all cyclists in the GLTS area. Models of participation were fitted at the local authority level using measurements of road traffic condition, population characteristics and modal use for journeys to work. Fitting 10 variables of these kinds explained over 80 per cent of variation in 6 of the 10 social groups. The proportions of people cycling in some social groups, including children, seem to vary quite strongly with simple indicators of traffic conditions. For possible application to areas other than London, a model which uses solely Census variables was fitted. The degree of variance explained exceeded 50 per cent in 5 of the 10 population groups. The proportion of workers who cycle to work, and car ownership, emerge as significant predictors of cycle use at the local authority level in these models. (A)

Want to know more about this project?