Ward level modelling of cycling to work

Examination of the 1981 census data showed significant differences in the cycle share of the work trip modal split. This paper describes work undertaken to develop a model, which might help with an explanation of this variation, for wards in a metropolitan area. The dependent variable was defined as ‘the proportion of the resident workforce in a ward that cycles to work’ (PCYJTW). For the study sample the defined dependent variable PCYJTW did not appear to be normally distributed. The model accepted as the preferred model employed the square root transformation and included three explanatory variables, representing traffic (which might be related to perceptions of danger associated with cycling to work), hilliness and car ownership. Several points gave rise to doubts that the selected model was causal, although it appeared to be statistically acceptable.