Habit and physical activity: Moderation and mediation studies in older adults

It is well-established that for older adults a physically active lifestyle is highly beneficial for health. Public health guidelines state that all adults should be at least moderately physically active for at least 30 minutes per day on at least five days per week. Unfortunately, a large proportion of older adults do not meet this physical activity (PA) recommendation. These older adults are insufficiently physically active to obtain the health benefits associated with regular and sufficient PA. Promoting PA in older adults is therefore of major relevance. Development of effective interventions to promote PA relies on insight into the determinants of PA and their working mechanisms. A determinant that merits examination in this regard is habit. The potential importance of habit lies in its close ties to PA maintenance, which, in turn, is essential for obtaining many health benefits. Although research has already revealed that PA has a habitual component, the longitudinal relationship between habit and PA can still be unraveled in more depth.The aim of this thesis is to contribute to a more comprehensive understanding of habit’s relation to PA. This understanding may inform intervention development. Chapter 1 describes the concept of habit, provides a brief history of habit from a psychological perspective, and defines PA for all studies in this thesis as the number of days per week on which a participant was at least moderately physically active for at least 30 minutes. Furthermore, this chapter describes that in health psychology for years the dominant approach has been to consider health behavior the result of a conscious, explicit, rational decision making process. This approach is known as the social cognitive approach. However, a more complete account of behavior can be achieved when in addition to explicit cognitions, such as intentions and attitudes, implicit processes, such as habits, are recognized. Models that account for both explicit, conscious influences and implicit, unconscious processes on behavior are called dual-process models. The first two studies of this thesis are concerned with PA as the outcome of a dual-process. The third and fourth study of this thesis test several mediation effects in which habit and PA play several roles. Chapter 1 specifies the research questions of the studies in this thesis in detail. All studies are conducted in older adults, aged 50 years or older. Chapter 2 presents a study that examined whether habit strength moderates the intention-physical activity (PA) relationship within the framework of the theory of planned behavior (TPB) and the attitude-social influences-efficacy (ASE) model. A longitudinal design with two measurement points was applied. Three PA habit groups were composed: a low, a medium, and a high PA habit group. The results of structural equation modeling (SEM) analyses showed that intention only significantly determined PA in the low and medium PA habit groups, implying that PA was not intentional at high levels of habit strength. Habit operated as a moderator of the intention-PA relationship. It is therefore recommended to add habit to the TPB/ASE model. This would transform this social cognitive model into a model that has characteristics of a dual-process model. As strong habits may limit the potential to change PA intentionally, only applying persuasive messaging as an intervention strategy may not suffice, and additional intervention strategies are needed for strongly habitual, but insufficiently active older adults. Chapter 3 shows a study that targeted the question why prior PA is a good predictor of later PA, even after TPB/ASE variables have been taken into account. This question is known as the residual variance problem. It has been suggested that habit operates as a mediator between prior and later behavior and as such forms at least a partial solution to the residual variance problem. In other words, prior behavior is proposed to exert its influence on later behavior through habit. This proposition was tested using a longitudinal design with four measurement points. Path analyses showed that habit indeed significantly mediated the relationship between prior and later PA, after TPB/ASE variables were taken into account. This result indicates that habit is a partial solution to the question why prior PA is a good predictor of later PA. It is recommended to incorporate habit into the TPB/ASE model. Chapter 2 and 3 both present studies that recommend to add habit to the TPB/ASE model. In these two chapters habit appears in two different roles: as a moderator that is concurrently assessed with intention, and as a post-intentional predictor of PA. Both roles not only indicate that habits have to be taken into account when explaining and predicting PA and when designing PA interventions based on the TPB/ASE model, but also that habit is a relevant construct in different phases that precede PA. The studies indicate that there is reason to consider incorporating habit into the TPB/ASE model in both roles. Both studies modeled habit as a variable that precedes PA. Additional insight into habit’s relation to PA can be achieved by modeling PA as a variable that precedes habit. In both Chapter 4 and 5 PA appears as a variable that affects habit. Chapter 4 reports on a study that applied a cross-lagged panel design to examine whether habit mediates the relationship between prior and later PA and whether PA simultaneously mediates the relationship between prior and later habit. The hypothesis that both mediation effects occur simultaneously was tested in two independent samples of older adults. While a significant PA-habit-PA path would support the implicit assumption underlying many PA interventions that PA sustains over time through habit, a significant habit-PA-habit path would indicate that PA interventions could benefitfrom incorporating habit formation strategies. The results of SEM analyses were not unambiguous. Indications for the existence of both hypothesized mediation effects were found, but no clear, unequivocal pattern appeared. Somewhat more support was found for the PA-habit-PA path than for the habit-PA-habit path. More research is needed to draw more definitive conclusions. Chapter 5 presents a study that modeled habit as an outcome variable. Habit formation has been proposed as a way to ensure long-term maintenance of PA and, as such, as a desired outcome of PA interventions. Intention, action planning (AP), and PA are suggested to be determinants of PA habits. It is, however, largely unknown how they determine PA habits. In two independent samples of older adults, the hypothesis was tested that the relationship between intention and habit is mediated by AP and/or PA. A four-wave longitudinal design was used. SEM analyses showed significant intention-PA habitpaths and nonsignificant intention-AP-habit and intention-AP-PA-habit paths in both samples. Thus, the relationship between intention and PA habit was mediated by PA and intention was neither associated with habit via AP as a single mediator, nor via AP and PA as sequential mediators. Possible conditions under which intention-AP-habitpaths and intention-AP-PA-habit paths exist are discussed in detail. Chapter 6 provides a summary and discussion of the main findings of the studies in this thesis, discusses methodological issues, and indicates implications for practice and directions for future research. A major strength of the studies in this thesis is the study population of older adults. Older adults are an underrepresented population in habit research. In light of both the rapid growth of this population and the large proportion of insufficiently physically active older adults, it is relevant to target older adults in PA habit research. Other major strengths are the large number of participants, the longitudinal design with four measurement points, and the use of two datasets in two studies. Limitations include the considerable and selective dropout and the measurement of PA by self-reports. The strengths and limitations should be taken into account when interpreting the findings of this thesis. Important implications for practice are that existing habits have to be taken into account when developing PA promotion interventions for older adults based on the TPB/ASE model, and that it maybe worthwhile to incorporate habit formation strategies, such as the use of reminders and self-monitoring, into interventions to improve the translation of PA into PA habits. An important direction for future research consists of close replications of the studies in this thesis as well as replications with variations in study population, time intervals, and target behavior. Other directions for future research include conducting experimental tests of the efficacy of habit formation strategies, gaining insight into the process of habit formation for different age groups, and testing the stability and development of habit over time. To conclude, this thesis shows that habit and PA are longitudinally connected with each other in several ways; there is continuous, reciprocal influence between habit and PA. Habits have to be taken into account when explaining, predicting, influencing, and maintaining PA. Being sufficiently physically active is associated with many health benefits. Through their influence on PA, habits for sufficient PA contribute to obtaining these health benefits.

The studies in this thesis were funded by the Netherlands Organization for Health Research and Development (Chapter 2-5: ZonMW, 6100.0003; Chapter 4-5: ZonMW, 200110006).