Comparison of model properties.

Electrical bicycle ownership rates are growing rapidly. Despite differences to conventional cycling, the two types of bicycles are generally not differentiated in travel demand modelling practice. This article analyses the choices to own electric and conventional bicycles in Germany at the personal level. We use data from the “Mobility in Germany” survey and other sources and estimate both a nested logit model and a multivariate probit model. While the average gradient of terrain near the residence has an expected, strong negative influence on the ownership of conventional bicycles, electric bicycle ownership is much less negatively affected. The effect of socio-demographic variables is largely in line with that of the existing literature. A negative correlation of the error terms in the probit model indicates a substitutive relationship between the two ownership decisions. The high nest parameter value in the nested logit model indicates that the decision to own a conventional bicycle is secondary to the decision to own an electric bicycle. The results contribute to a better understanding of the motivations for or against bicycle ownership and create a basis for better consideration of electrical bicycle traffic in transport models.