Search

The rising popularity of physical activity tracking devices and online social networks has seen them touted as potential tools to positively affect physical inactivity. These systems have found favour with a broad spectrum of users, from serious, competitive athletes to people trying to become more active, lose weight and boost wellbeing. To help determine their suitability to potentially assist with community health issues it's vital that we learn more about how they are used in practice, the exercise outcomes that accrue and how users relate with one another online. We need to understand how these systems are designed to target specific, positive exercise behaviours and determine how they might accommodate health behaviour change theories to act as agents of health change. The studies that comprise this thesis examine features of a population of users of an exercise tracking system known as movescount.com and the role the characteristics of these people and the design of the system may play in exercise outcomes and online interactions. The first study involved an initial demographic and anthropometric analysis of the user population of the Suunto Movescount database of twenty thousand users. The results revealed users are predominantly male (89%) and list running as their main exercise (28%) with only 16% of them operating an external social media account. They were also quite lean with an average BMI (Body Mass Index) less than 25 and fit with a mean system-based Fitness Index Score of 5.7 (showing above average fitness). A classification and regression tree analysis was performed to find a group of attributes that classified users as persistent or non persistent. This work showed the fitter users used the system more frequently and of the least fit, those with an active Twitter account persisted with exercise for longer than their equivalent peers. Users with lower BMI scores and greater numbers of followers tended to login to the activity tracking system more often. An examination of the system's equivalent to Facebook likes function revealed 75% of users positively self-affirmed their own exercise. Those that published their moves not only to the activity tracking system but also to Twitter persisted in using the system for longer, logging into the tracking system 13.2 more times and uploading their moves 5.1 times more than the average user. The second study focused on the persuasive elements of the Suunto system. An online survey was administered to a subset of users of the system using questions to determine exercise status and device use. The survey incorporated a validated scale based on the Behaviour Change Support System (BCSS) theory, which was modified by incorporating two other standardised scales that measure relatedness to others in offline physical activity (ROPAS), (Lehto et al., 2012) and the online sociability sub scale of the Brief Test of Online Behaviour (BTOB) by (Johnson & Kulpa, 2007. A factorial analysis of the results found the intention to continue using the system was determined by the users' perceptions of the effectiveness of the system, the effort required to use the system, the credibility of the system and the social support offered by the system. It was also revealed that any individual who perceives high levels of relatedness to others during physical activity in the real world and identifies strongly with the online community of fellow digital exercise system users will likely feel they receive high levels of social support from the system, thereby positively affecting their use of the system. Structural equation modelling shows that an increase in age and being single brings statistically significantly lower ROPAS. This implies that as users age and live as singles their relatedness to others in physical activity declines and with it a propensity to engage with the systems social support functions may follow. The third study engaged a panel of five experts experienced in the theoretical aspects and practical application of BCSS to the design, development and deployment of health behaviour change systems to complete an independent evaluation of the Suunto movescount system. They found that the system rates strongly for primary task support and social support but only satisfactory for systems dialogue and credibility. The panellists also indicated it was deficient in Reminders and Suggestions, design cues regarded as crucial for persuasiveness in apps. Further, the lack of adequate Simulation and Rehearsal functions may impinge on the planning and execution of user exercise; (Holmes & Calmels, 2008). The panel found the system lacked in ideal levels of praise and reward for user exercise efforts, a feature considered important in health apps design; (Baranowski, Thompson, Buday, Lu, & Baranowski, 2010; Thompson et al., 2010) A recommendation from these findings is that product development teams engage with behavioural scientists during the development phase to make best use of health behaviour change and persuasive systems practices. The final study investigated if individuals who tweet their exercise efforts from the movescount.com system exercised differently according to their online social influence. A quantitative analysis of a sample of such users was conducted. This determined if there was anything out of the ordinary by way of exercise volume, frequency and type and whether or not an independent measure of their online social influence furbished by the KLOUT.com service could reveal if those who scored highly There was no relationship between exercise session duration or intensity and online social influence. The thesis identifies an activity tracking system as an effective technology for persuading individuals to maintain exercise behaviour. It has identified that those users that score highly on a measure of relatedness to others in physical activity in the real world and identify strongly with the online community of peers will indicate they receive high levels of social support from the system and are more likely to continue using the system for longer. These findings may have implications for designers of activity tracking systems through the inclusion and exclusion of social networking functions by user profile as determined by a programmatic operationalization of the validated scale created by this thesis.