LTPA is a behavior that involves different types of activities (e.g., group, individual, recreational and competitive activities), which occur in different social contexts for varied lengths of time and with varied levels of physiological demands. Because of this scenario, it was decided to investigate various features of physical activity in order to understand the characteristics of the relationship between social support and LTPA better. This study examined the association of social support dimensions (i.e., material, emotional/information, affective and positive social interaction) with four LTPA outcomes (engagement, maintenance, LTPA type, and time spent on LTPA). Our results suggest that the influence of social support on LTPA depends on the social support dimension, LTPA outcomes and the group evaluated (those recently engaged or those who maintain LTPA). It is thus plausible that there are different pathways linking social support and LTPA. In our view, the material and emotional/information dimensions might be directly linked with LTPA because they relate the availability of physical activity resources and exposure to health information, respectively. On the other hand, the positive social interaction dimension might be linked to LTPA by providing motivation and self-efficacy. The role of self-efficacy as a mediator of the relationship between social support and health-related behavior has been demonstrated previously in the physical activity literature [12, 34]. Moreover, several theories attempt to explain how protective behaviors are initiated or maintained. The main idea of these theories is that motivation toward protection results from a perceived threat and the desire to avoid the potential negative outcome. In other words, the motivation is related to the health and aesthetic benefits that a physical activity could provide. Thus, the positive social interaction dimension can be linked to this pathway, because it involves informal social control through norms and attitudes. It could then be related to higher or lower levels of physical activity, depending on the context established by the social network providing the social support. Our results show that positive social interaction in the form of material and emotional/information supports was related to higher levels of LTPA, suggesting that members of the study population were surrounded by social networks that tend to support the practice of physical activity. On the other hand, we did not find an association between dimensions of social support and LTPA based on the filter question (whether any physical activity had been performed in the previous two weeks), a negative finding that could have resulted from the generic phrasing of the LTPA question. This finding emphasizes the importance of using more specific LTPA variables. Also, there is weak evidence of the affective dimension's influencing LTPA; only in the relationship between this dimension and LTPA type did we find a significant association. These findings may reflect the characteristics of the dimension, in that affective support may exert a more indirect influence on LTPA than the other dimensions.
In the engagement group results, all dimensions of social support are related to engagement in group activities, but not in individual activities. These results are interesting because engagement in group activities is often more difficult for the following reasons: first, accessing specific materials and locations for group activities, which could be related to material and emotional/information dimensions of social support, are the first practical steps to beginning a group activity; and, second, knowing or learning certain basic rules and techniques for the specific physical activity often requires instrumental support. However, some group leisure-time physical activities are so traditional that they are intrinsically familiar (e.g., soccer in Brazil, basketball in the United States). Finally, arranging the time for all participants to perform the activity could be a barrier. Thus, it is plausible that individuals with higher levels of social support are more likely to surpass all these barriers and join in a group activity than are others with low levels of social support. The results for time spent on LTPA are less striking than for LTPA type, although individuals with high levels of the emotional/information and positive social interaction dimensions of social support are more likely to perform more than four hours per week, as compared with the others who performed only a maximum of 2 hours per week. These findings indicate two different modes of social support: first, the influence of the emotional/information dimension on the time spent on LTPA is related to the exposure to health information that could improve knowledge of the benefits of physical activity . Second, the social positive interaction dimension significantly increases the possibility that an individual will be in contact with individuals with whom to engage in leisure activities, including physical activities.
In the maintenance group, only the material dimension influenced LTPA type, and the emotional/information and social positive interaction dimensions were related to time spent on LTPA. These findings suggest that, among individuals still involved in physical activity after two years of follow-up (between 1999 and 2001), only practical aspects, such as access to appropriate materials or locations, were important to their continuing or engaging in group activities. In other words, interactions with individuals represented by the positive social interaction dimension could positively influence motivation to perform, and the sense of confidence in performing, a physical activity, which would, consequently, increase the amount of time spent on LTPA. As self-efficacy theory suggests, the information and feedback that an individual gains from performing an activity and the belief in their enhanced ability to perform the activity could be related to maintenance of the activity and the time spent performing it . In addition, the maintenance group could be exposed to basic information about physical activity (e.g., time and intensity) and might perform the activities based on this information. It could be that middle and high levels of the emotional/information dimension are related to being involved in LTPA for more than three hours per week, a level that is closer to current health recommendations.
Overall, the results did not show any simple dose-response effect relating levels of social support dimensions and aspects of LTPA. Furthermore, an intermediate level of positive social interaction seems to be more important than the highest level in relation to time spent on LTPA. These findings suggest that the intermediate level of social support may be sufficient to influence LTPA and that the highest level of social support may not yield any additional impact on LTPA. It may also be that, to some extent, the highest level of support reflects the downsides of social relationships . It is plausible, for instance, that highly supportive relationships sometimes provide information that discourages rather than promoting LTPA.
Despite the fact that comparisons between engagement in, and maintenance of, LTPA were not the focus of this study, it is notable that the influence of social support differs between the engagement and maintenance situations, suggesting that social support has different impacts on these groups. Our findings suggest that social support is more important to engagement in, than to maintenance of, physical activity. Nevertheless, a previous study  suggests that social support is equally important in both situations.
Although we did not find studies using time and type of LTFA as the main outcomes to investigate the potential influence of social support, our results are in line with previous work which observed associations between social support and LTPA, either in general population-based studies [11, 38] or in specific subgroups [10, 22]. For example, one study  found that instrumental church-based social support helped initiation of physical activity in a rural population.
Some limitations of our study should be noted. The use of self-reporting to measure LTPA and the use of a social support instrument that did not focus on LTPA may have limited the scope for comparison with other studies' findings. On the other hand, with these measurement strategies, we generated helpful LTPA outcome variables and investigated the role of all social support dimensions on LTPA. Second, time spent on LTPA, as reported in the questionnaire, may have been overestimated. However, the strategy of individuals filling in the information about time spent on LTPA separated by activity and session probably minimized this problem. Third, this is a specific occupational cohort of public employees in Rio de Janeiro, probably with higher levels of LTPA, and it is uncertain how far the findings of this study can be generalized to the overall population of Brazil or to other occupational groups and countries. Fourth, because the study design was based on access to LTPA data at only two points in time, it was not possible to evaluate for possible changes in LTPA that may have occurred during the follow-up period. Fifth, some models returned large confidence intervals of the effect measure evaluated in the study, probably due to missing values. To evaluate the impact of this problem, we performed models based on multiple data imputations and a sensitivity analysis which found similar results. Finally, another possible criticism of the study is that engagement in/maintenance of LTPA may result from health campaigns promoted by the university. However, the fact that none took place during the period covered by the study makes our results even more robust.