Intention, Habit and Health Behavior as Dynamic Constructs: A Multilevel Analysis of Longitudinal Data

Introduction
Dual-process theories suggest that health behavior is governed by both reflective and automatic processes. Reflective processes involve conscious planning such as intention, whereas automatic processes operate without conscious awareness and are driven by learned associations, e.g., habit. Research on intention and habit hast generally relied on cross-sectional methods. Evidently, these methods are inherently incapable of capturing variations over time. The analysis of longitudinal data is required to separate between-person differences and within-person variation. Therefore, this study examines how temporal dynamics in intention and habit influence the development of health behavior over time.

Methods
The sample comprised 182 participants (61% female and mean age 24 years) from weekly sports classes at two German universities, monitored over 13 weeks for class attendance. Habit and intention were measured weekly using one automaticity item from the automaticity subscale of the Self-Report Habit Index (SRHI) for habit, and one intention item measuring the intention to participate in the next session. Thus, intention and habit were only measured for attendees, reducing the total number of observations from 2224 to 1175 for the analyses including habit and intention.
... mehr

Results
Three multilevel models were constructed to investigate the influence of time, intention and habit on attendance, accounting for the hierarchical structure of the data, (1) time (between-person), (2) time and intention (within-person), and (3) time, intention and habit (within-person). There was a 24 % decrease in the odds of attendance per week (OR = 0.76, p < .001) in the first model, reflecting a stark decline in attendance over time. In the subsequent lag-1 analysis, there was a 22% increase in the odds of attendance for intention, (OR 1.22, p < .001) reflecting a significant within-person effect of intention, however, there was no within-person effect of habit in the final model.

Discussion
The study highlights the importance of longitudinal designs and multilevel methods for capturing the complex dynamics of intention and habit. By modeling time and individual predictors, within-person effects (week-to-week changes) and between-person differences can be disentangled. The study shows that attendance decreases even amongst healthy students, thus highlighting the importance of understanding attendance and health behavior maintenance in rehabilitation.