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

Hintergrund und Zielstellung: Dual-process theories suggest that health behavior is governed by both reflective and automatic processes. Reflective (explicit) processes involve conscious planning, whereas automatic (implicit) processes operate without conscious awareness and are driven by learned associations (Hagger, 2016). Intention represents the reflective processes, while habit reflects the automatic processes involved in behavior regulation. Previous research shows that intention and habit predict health behavior, that habit increases with repeated behavior, and may reduce the influence of intention on health behavior (Gardner et al., 2020). Both intention and habit are continuously influenced by situational and temporal factors. However, research on intention and habit tends to rely 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 (Knapova et al., 2024). This study examines how temporal dynamics in intention and habit influence the development of health behavior. ... mehrSpecifically, it seeks to understand how intention and habit evolve and influence one another over time, emphasizing the importance of within-person variance in the predictors of health behavior.

Methoden: The sample consisted of 182 participants (61 % female and mean age 24 years) attending weekly sports classes at two German universities. The participants were monitored over a 13- week period, with class attendance as the dependent variable. Habit and Intention were measured weekly with attending participants answering one automaticity item derived from the automaticity subscale of the Self-Report Habit Index (SRHI) (Gardner et al., 2012.) for habit, and one intention item measuring the intention to participate in the next session (Finne et al., 2019). Thus, intention and habit were measured only if a participant attended the session.

Ergebnisse: 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 decrease of 0.27 in the log odds of attendance per week (SE = 0.02, z = -17.72, p < .001) in the first model, reflecting a stark decline in attendance over time. Therefore, although 182 participants were included in every analysis, the total number of observations was reduced from 2224 to 1175 when intention and habit were included (model 2 and 3). Unsurprisingly then, there was no significant effect of intention, habit, or their interaction in the second and third model.

Diskussion und Fazit: The results showed a decline om attendance over time, unfortunately leading to systematically missing data. Thus, there was no effect of intention or habit on attendance among participants who regularly attended the sports classes. 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. Take-Home-Message These findings show how multilevel models can enable a nuanced understanding of temporal and individual variability, emphasizing the need for longitudinal designs in future research.