In an era where activities are inevitably linked with the use of technology, such as the use of computers for online shopping and televisions as a source of entertainment, in order to understand the energy implications of transport policies it is important not only to understand the activity-travel patterns of individuals but also their choice of technology holdings. In this paper, we propose to model the in-home and out-of-home non-work activities undertaken by individuals jointly with their choice of technology holdings. The overall objectives of this research are two-fold. First, such a model examines the potential substitution effects between the in-home and out-of-home activities of an individual. Faced with policies such as increased fuel prices, individuals may choose to substitute some of their out-of-home activities with equivalent in-home activities such as, for instance, online shopping instead of travelling to the supermarket. However, people have varying priorities and preferences for technology use. Accordingly, while person A may prefer to substitute shopping trips with online shopping, person B may prefer a personal shopping experience involving travel to the store but may substitute visits to the theatre with watching television at home. Second, one would expect the choice of technology holdings to be closely correlated with individual (or household) preferences and priorities. Technology not only has the potential to enhance the quality of the in-home activity but can also be indication of an individual's intrinsic lifestyle preferences. Modelling technology holdings jointly with in-home and out-of-home activity participation could therefore capture a preference endogeneity that would otherwise be considered to be 'unobserved'.

Few data sources, however, contain all the data required for such a study. Although most activity-based travel surveys contain complete activity-travel data for the surveyed respondents, the data on household technologies is typically limited to the availability of a home computer. We therefore use an ensemble of data sources including the UK National Travel Survey data, the UK Time Use Survey data and the British Household Panel Survey data, and are faced with the task of combining these heterogeneous data sources. There is a vast body of literature on this topic ranging from the pooling of revealed and stated preference data, to the combination of sequences in sociological and genomic research. A variety of techniques are available such as likelihood partitioning, bayesian graphical/network modelling, multiple imputation, and synthetic population generation using iterative proportional fitting, as well as ad-hoc procedures such as clustering followed by sampling. In this study, we first propose to compare these techniques and their effects on the model specifications and results.

Model estimation will be undertaken in two parts. First the non-work activity participation of individuals on a given day will be modelled treating their technology holdings as exogenous. This is essentially a multiple discrete choice problem with a choice of in-home or out-of-home location for each of the non-work activity categories – shopping, personal business and leisure. We propose to use a multi-category choice modelling specification by defining bundled choices. Finally, a joint model of activity participation and technology holdings will be developed to determine the degree of endogeneity in these choices.