Space, time, activity and human error: Using space-time constraints to interrogate the degree of uncertainty in survey-based movement datasets

Tourist activity is generally frenetic even while seemingly being relaxed. A significant change has been the rise in free, independent travellers who choose to tour autonomously and visit multiple destinations to their own schedules. This development has had major ramifications, impacting on local environments and communities by stimulating their economies but simultaneously demanding new facilities, displacing certain activities, and transmitting ideas and even disease as tourists contacting with their hosts becomes wider and more intense. Such tourism is quintessentially tied up with a dynamic geography of movement that generates demand and supply at different spatial scales. A growing recognition of these outcomes has highlighted the significance of movement data as a resource for understanding many aspects of human and animal activity and their geographies. Consequently, research interest has accelerated on the back of enhanced capabilities for tracking individual entities' movements, typically with GPS sensors that collect individual time-tagged locational data cheaply and accurately. Prior to this, most movement studies used a paper-based survey methodology for data capture which was reliant on respondents' recall of movement or the keeping of a diary. Unlike the GPS, this process permitted data capture which is enriched by information on the respondent's profile, and ongoing activity, time use, and attitude, a distinction which continues to validate this methodology in a number of contexts. Legacy datasets gathered using surveys are known to have (non fatal) sources of inaccurate or incomplete responses, which in general have been documented only to a limited degree. This paper is concerned with using GIS technologies to more fully interrogate a case study database (tourists travel survey) so as to identify: (i) the level of uncertainty in given responses from individuals, (ii) the pattern of missing data and (iii) the degree to which such datasets can be enhanced by models using concepts found in time-geography.