The main objective of this study was to identify a number of moderators of the intention-behaviour and PBC-behaviour relationships regarding leisure-time physical activity. The moderators of the intention-behaviour and PBC-behaviour relationships were age and annual income. More importantly, however, only one of these moderators remained significant in the final model, namely annual income as a moderator of the PBC-behaviour relationship. In the domain of physical activity, several studies have reported positive associations between physical activity and annual income (see Trost, Owen, Baumen, Sallis and Brown [35
] for review). However, to our knowledge, this is the first study to report moderating effects of annual income on the PBC-behaviour relationship.
The PBC-behaviour relationship was higher for individuals with a higher income compared to those with a lower income. Ajzen [4
] stated that PBC provides an accurate prediction of behaviour only when individuals have realistic perceptions of control over a given behaviour, meaning when individuals' perception of control matches actual control over behaviour. This explanation is congruent with Sheeran, Trafimow and Armitage [15
], who for exercise behaviour observed a significant difference between PBC and a proxy measure of actual control among 73% of their study sample (N
= 226). They reported that when PBC was realistic, PBC explained twice as much variance in behaviour compared to unrealistic PBC. Thus, in the context of the present study, it can be suggested that individuals who have a higher income are better able to evaluate their true control over behaviour. Consequently, their expression of control is likely better aligned with true control, either because they face fewer barriers or have the ability to overcome such barriers to physical activity.
Although the other moderators did not reach significance in the final model, they were nonetheless significant moderators when considered individually. In this regard, annual income was also a moderator of the intention-behaviour relationship. Again, it can be suggested that individuals with higher income face fewer barriers to leisure-time physical activity and have better resources than individuals with a lower annual income. Thus, individuals with higher annual income present better intention-behaviour consistency than individuals with lower annual income.
Age was another factor found to moderate the intention-behaviour and PBC-behaviour relationships. Two meta-analytic reviews of the TPB in the field of physical activity have tested age as a moderator of the intention-behaviour relationship. Hagger, Chatzisarantis and Biddle [6
] found that younger samples (adolescent and college, aged under 25, based on mean age study samples) showed a significantly weaker intention-behaviour relationship compared to older samples (aged 25 or older, based on mean age study samples). Similarly, Downs and Hausenblas [7
] reported a smaller intention-behaviour relationship for samples of children/adolescents (aged 8 to 17) compared to samples of young (aged 18 to 25), middle-age (aged 26 to 64) and older (aged 65 or older) adults. Obviously, our results are in agreement with these observations. Among the potential explanations for these results, it may be suggested that older people (i.e., 50 to 55 years in the present study) are more likely to have an established routine (lifestyle) and, consequently, intention is better aligned with their behaviour. Young adults are more likely to deal with unstable living situations (e.g., entering the workplace full time or a new school, meeting new friends, leaving home, living with a new life partner) and may experience more disruption in life, thereby resulting in a lower intention-behaviour relationship. However, given that Downs and Hausenblas [7
] have also observed that middle-age adults (aged 26 to 64) had a weaker intention-behaviour relationship compared to older adults (aged 65 or older); additional studies on the moderating role of age are needed.
Concerning the PBC-behaviour relationship, being older (i.e., 50 to 55 years) was associated with greater PBC-behaviour consistency. This observation is similar to the results reported by Notani [30
], who showed that the PBC-behaviour relationship was better among non-student samples (older individuals) compared to student samples (younger individuals). To our knowledge, this moderating role of the PBC-behaviour relationship has not been reported previously in the specific field of exercise/physical activity. The potential explanation for this association was discussed earlier with respect to annual income, but taken together, this factor reinforces the assumption that older individuals have a better evaluation of the true control they have on the regular practice of leisure-time physical activity.
From a practical point of view, the results of this study provide useful information to guide the promotion of physical activity. First, intention was found to be a significant determinant of behaviour. This is congruent with Sheeran's [11
] meta-analysis showing that on average, the explained variance between intention and behaviour corresponded to a "large" effect size [8
]. Thus, education efforts aimed at increasing intention to practice leisure-time physical activities remain necessary to favour the formation of a positive intention and the regular practice of leisure-time physical activity. This, however, should be done in parallel to other strategies concerned with lowering barriers to behaviour adoption and increasing control over adoption of an active lifestyle. Indeed, among specific sub-groups, other actions will be necessary to strengthen the effectiveness of physical activity promotion interventions. For instance, concerning individuals with a lower income, the interventions, in addition to motivational considerations, should be concerned with actual control. Alternatives to usual supervised physical activity programs and facilities should be offered. The promotion of low cost physical activities or "free" access to physical activity programs should be considered by public health authorities. Obviously, each segment of the population would benefit from specific interventions unlikely to suit all individuals. However, the promotion of leisure-time physical activity should combine strategies aimed at both motivational and control dimensions.
To our knowledge, this is the first study to test social structural factors such as educational level and annual income as moderators of the intention-behaviour and PBC-behaviour relationships. To date, most tested moderators were either cognitions or age and gender. However, given the increasing interest towards the contribution of social structural factors (as well as environmental variables), this timely information is necessary. Indeed, much remains to be learned in order to increase our understanding of the conditions and contexts explaining why some individuals are successful in translating their intentions into action, while others are not successful in this task.
Some limitations in the present study should be acknowledged. First, our study was conducted among a group of well-educated volunteers. Such individuals are likely to have higher intentions and levels of physical activity compared to the general population. Secondly, the non-optimal distribution of intention/PBC and some moderator variables (dichotomous variables like FHO and educational level) may have compromised the power of detection of some moderators, making detection of other moderators more powerful [45
]. Thirdly, one of the biological variables, FHO, was based on the subjective evaluation of the participants. This may have compromised its true role as a moderator. Nonetheless, given that two studies reported a good reliability of an individual's report of family weight and height [46
] we have confidence that our results were not affected by this method of measurement. Fourth, only leisure-time physical activity was studied. It is possible that other aspects of physical activity (e.g., transport) would have identified other moderators. Finally, the measure of physical activity was self-reported; however, we used a validated physical activity questionnaire [42