Summary and conclusions
This paper presented the use of consumer panel data and hazard based duration models to explore the potential impacts of online shopping separately on shopping trip frequency and overall shopping activity patterns. Results provide new insights to substitution or complementarity question. Additionally, methods presented can be used for predicting next online order or shopping trip, which is crucial for activity generation models used for travel demand predictions and also for delivery operations. Our study has a number of novel components in comparison with past work that focused on understanding the implications of wider use of ICT on personal travel.
First, while transport researchers have used hazard based duration models to analyse inter-shopping duration in the context of multi-day activity generation, they have not considered online shopping activity and its effects on overall activity generation. Our study accounts for the influence of online activity on temporal patterns of shopping activities.
Second, our analyses are based on a dataset of unprecedented richness in transport research with respect to capturing both online and physical shopping activities. In consumer panels, online shopping activity data are collected in an episode-based manner, rather than through retrospective questionnaires where respondents report how frequently they shop using a points scale. This enables development of more advanced models to help us better understand changes in behaviour triggered by ICT use. Additionally, shopping basket characteristics, likely to affect shopping behaviour yet often neglected in transport literature, are available and used in empirical estimations. Importantly, market research companies operate consumer panels in many countries and make it available to researchers. This is crucial for the generalisability of our proposed method.
