JMIR Publications

ABSTRACT

Background:

While physical activity can mitigate disease trajectories, and improve and sustain mental health, many people have become less physically active during the COVID-19 pandemic. Personal information technology, such as activity trackers and chatbots, can technically converse with people and possibly enhance their autonomous motivation to undertake physical activity. The literature suggests that for an effective design of such interactions, adopting Behavioural Change Techniques (BCTs) based on Self-Determination Theory (SDT) seems promising, but this remains untested.

Objective:

The objectives of our study are (1) to test whether autonomous motivation for walking can be increased when a chatbot in combination with an activity tracking smartphone application (app) is used, (2) to confirm the underlying theoretical mechanisms, and (3) to evaluate the effectiveness of various BCT implementations.

Methods:

We employed a 2x2x3 factorial field experiment, using 12 variations of a chatbot which differed in three BCTs: goal setting, experimenting, and action planning. In total, 102 participants used a variation of the chatbot together with the Google Fit app over the course of three weeks. Each week, participants were asked to have a conversation with the chatbot and to complete a questionnaire capturing their perceived app/chatbot support, need-satisfaction, and physical activity levels. Motivation was measured before and after the three-week period.

Results:

On average, across all variations of the chatbot, participants reported significant increases in autonomous motivation (P<.001). Motivation was associated with need-satisfaction (P<.001) and need-satisfaction was associated with perceived app/chatbot support (P=.002). In terms of the different BCT implementations no significant differences were found. While many participants (49%) would have preferred to interact with a human instead of the chatbot, 46% of the participants stated that the chatbot helped them to become more active, and 42% of the participants decided to keep using the chatbot for an additional week. Furthermore, a majority thought that a more advanced chatbot could be very helpful.

Conclusions:

The results provide evidence that a chatbot in combination with a physical activity tracking app such as Google Fit can increase autonomous motivation by supporting the needs of competence and autonomy. Our study also clarifies a need to further study how the corresponding Behavioural Change Techniques are best implemented, and how other BCTs could be studied.