Background: Digital smartphone messaging can be used to promote physical activity to large populations with limited cost. It is not clear which psychological constructs should be targeted by digital messages to promote physical activity. This gap presents a challenge for developing optimal content for digital messaging interventions.
Objective: The aim of this study is to compare affectively framed and social cognitively framed messages on subsequent changes in physical activity using dynamical modeling techniques.
Methods: We conducted a secondary analysis of data collected from a digital messaging intervention in insufficiently active young adults (18-29 years) recruited between April 2019 and July 2020 who wore a Fitbit smartwatch for 6 months. Participants received 0 to 6 messages at random per day across the intervention period. Messages were drawn from 3 content libraries: affectively framed, social cognitively framed, or inspirational quotes. Person-specific dynamical models were identified, and model features of impulse response and cumulative step response were extracted for comparison. Two-way repeated-measures ANOVAs evaluated the main effects and interaction of message type and day type on model features. This early-phase work with novel dynamic features may have been underpowered to detect differences between message types so results were interpreted descriptively.
Results: Messages (n=20,689) were paired with valid physical activity monitoring data from 45 participants for analysis. Received messages were distributed as 40% affective (8299/20,689 messages), 39% social-cognitive (8187/20,689 messages), and 20% inspirational quotes (4219/20,689 messages). There were no statistically significant main effects for message type when evaluating the steady state of step responses. Participants demonstrated heterogeneity in intervention response: some had their strongest responses to affectively framed messages, some had their strongest responses to social cognitively framed messages, and some had their strongest responses to the inspirational quote messages.
Conclusions: No single type of digital message content universally promotes physical activity. Future work should evaluate the effects of multiple message types so that content can be continuously tuned based on person-specific responses to each message type.
Keywords: Fitbit; behavior change; dynamical model; exercise; fitness; fitness tracker; messaging; motivation; patient specific; patient-specific modeling; physical activity; psychological theory; tracking.
©Alexandra M Lee, Sahar Hojjatinia, Jimikaye B Courtney, Deborah Brunke-Reese, Sarah Hojjatinia, Constantino M Lagoa, David E Conroy. Originally published in JMIR Formative Research (https://formative.jmir.org), 21.04.2023.