Our study results showed that education, family income, self-rated health, intention, self-efficacy, perceived barriers, perceived health benefits, and facility availability were independently related to PA. Self-efficacy and intention were the strongest correlates and had the greatest effect on PA. Self-rated health, family income, perceived health benefits, and perceived barriers were also consistently associated with PA. In addition, the effects of perceived health benefits, education and family income were more salient to older people, whereas the influence of education was more important to women and the influence of perceived barriers was more salient to women and younger people. Furthermore, facility availability was more strongly associated with PA among people with a university degree than people with a lower education level.
This study found that higher self-efficacy was consistently related to higher PA across gender, age group, education level and family income level. This finding is in agreement with other studies on an array of populations [11
]. Confidence in personal ability to carry out a behaviour (i.e., self-efficacy) plays a central role in behaviour change and influences the direction, intensity and persistence of the behaviour [23
]. This focal belief is the foundation of human motivation and action. People who have higher PA self-efficacy will perceive fewer barriers to PA or be less influenced by them, be more likely to pursue perceived benefits of being physically active, and be more likely to enjoy PA [23
]. One study [24
] suggests that self-efficacy determines whether people translate perceived risk into a search for health information and whether they translate their knowledge into healthy behavioural action. The study [24
] also found that knowledge-behaviour correlations were greater among those with high self-efficacy, increased among those who raised their self-efficacy, and decreased among those who reduced their self-efficacy. Intervention studies also showed that enhancement or manipulation of perceived self-efficacy resulted in an increase in PA level or in adherence and maintenance of the exercise behaviour [16
]. On the other hand, engagement in PA can affect a person's self-efficacy [20
]. A prospective study found that participants who exercised more frequently during a 6-month structured program had a more positive exercise experience, which, in turn, enhanced their self-efficacy at program end, resulting in higher levels of exercise participation at 6- and 18-month follow-up [19
]. Self-efficacy is usually higher among men than among women and is positively related to socioeconomic status [28
Intention is another important independent correlate for PA in this study. Our finding of the positive influence of intention on PA participation corroborates those of other studies [29
]. Intention, an essentially proximal goal, would provide self-incentives and guides for health habits as well as help people to succeed by enlisting effort and guiding action [23
The strong effects of self-efficacy and intention on PA suggest that interventions designed to increase PA should target self-efficacy and intention. Self-efficacy can be influenced by reinforcement history, observational learning, and perceived exertion [33
]. Therefore, future research is needed to identify how those influences can be optimally incorporated into interventions that will increase people's beliefs on their ability and motivation/intention to be physically active.
Our results indicate that higher SES, including higher family income level and education level, is positively associated with PA, although the association between education and PA was significant among women but not among men. Many studies found a positive association between higher education or higher income and PA levels [6
]. People with higher education levels are more likely to have better general health, higher self-efficacy (due to stronger problem-solving and coping capacities arising from educational experience), more social support, and a greater capacity to seek, understand, and act on health messages that promote PA [6
]. In addition, people with higher family income levels usually have better health (due in part to better access to health care resources), have better access to PA facilities and opportunities, can choose and afford to live in a pleasant and activity-friendly environment, and have less barriers to PA [6
]. Our results of stronger associations with self-efficacy, perceived health benefits and facility availability among people with a university degree than among those with lower educations, as well as stronger associations with intention and perceived health benefits and a weaker/no association with perceived barriers among people in the category of highest family income level provide supports to the role of the SES on PA level. No significant association between education and PA among men could be because that men with lower education levels are more likely to have jobs of physical labour (therefore have higher occupational PA) than people with higher education levels [38
]; therefore, men with lower education levels may have similar total PA as those with higher education levels even though they may have lower recreational PA. It is also possible that, compared to men with lower education levels, women with lower education levels are more likely to have sedentary jobs such as clerks and secretaries [38
], therefore having an occupational PA level more similar to women with higher education levels.
This study also found that self-rated health was strongly and consistently related to PA across sex, age group and SES. Perceived poor health has been reported to be associated with lower PA level in other studies [22
]. A cross-sectional study of 16,230 respondents in the 15 member states of the European Union also observed a higher level of total PA associated with better self-related health across populations [41
]. However, one study of urban women indicated that self-rated health was not a significant correlate of leisure-time PA [42
Our finding of the positive correlation between perceived health benefits and PA level is in line with other studies [11
]. The results of two studies on samples of mainly males [21
] also support our finding. One possible explanation for the more salient effect of PA's health benefits on PA level in older people than in younger people is that older people usually have more health problems, therefore consider PA health benefits more important for them than younger people do. Younger people might consider other benefits more important in their decision to participate in PA, such as enjoyment, social interaction, improvement of self-esteem, better shape, increased attractiveness, and strength. Therefore, future studies assessing PA benefits should include not only health benefits but also other psychological benefits, while there is a need to include both benefits in the education message in developing interventions of promoting PA.
Perceived barriers as an important factor for PA participation have been demonstrated in many studies [12
]. That perceived personal barriers appear to be more important to women than to men might reflect the situation that women devote more of their time to their multiple responsibilities as workers, housekeepers, mothers and wives. The greater effect of barriers on PA in women than in men may be also because men may enjoy PA more than women, and men usually have higher self-efficacy for PA thus perceiving less barriers or less influenced by barriers [28
]. In addition, on average, men have higher levels of occupational PA than women [38
] while perceived barriers to PA are mainly related to recreational PA. One possible reason of a stronger effect of perceived barriers among people <65 years than among seniors might be that seniors are usually retired so they usually have more free time than younger people.
Both perceived and objectively measured physical environment factors were found to be positively related to PA level [7
]. Availability, accessibility, convenience of destinations and facilities as well as the general functionality of the neighbourhood (e.g., traffic condition, street lighting at night, unattended dogs and safety from crime) and aesthetics have been shown to be positively associated with PA level [7
]. A meta-analysis [10
] found a modest, yet significant association between the perceived physical environment and PA. Literature suggests that the built environment can affect people's decision for participating in PA by providing cues and opportunities for activities to occur [9
]. Some studies indicated that the physical environment also had an indirect effect on PA through self-efficacy [18
]. However, our study suggests that perceived facility availability was significantly associated with PA only among people with a university degree. This study assessed only availability, while accessible, convenient and safe facilities for PA might be more strongly associated with PA than availability [48
] because people would not use those available facilities if they are too expensive, not convenient and not safe. Therefore, accessibility, convenience and safety of PA facilities should be assessed in future research on the physical environment. Also, both perceived and objectively measured physical environment data should be included in the same studies.
Although we did not observe an independent effect of social support on PA, many studies have shown the importance of social support in promoting PA [11
]. Some studies have also shown that social support has an indirect effect on PA through self-efficacy [18
Limitations of our study should be considered when interpreting the results. First of all, our study was a cross-sectional design and causal inferences cannot be made because of the inability to determine temporal sequence. Prospective study designs should be considered in further research on these relationships in order to provide more insight on the question of the causal direction. Secondly, the response rate was low (51%) and there might be inherent differences between people who agreed to participate in the study and those who did not. However, earlier analyses showed no response rate bias [50
]. Furthermore, this response rate was similar to that of other PA surveys in other countries such as Australia, the US and the Netherlands [11
]. Another limitation common to population surveys was that PA measures were self-reported, where respondents may over-report their occasions or time spent in PA. The survey assessed total physical activity only, therefore, we could not examine the correlates for occupational, recreational and transport-related PA separately, whereas some factors are related to recreational and transport-related PA only. Future research should collect separate information on these types of PA in order to understand the differential effect of various factors on specific types of PA. Also, our current analysis did not assess mediation between different factors (individual factors such as self-efficacy and intention may mediate the influences of social factors, environmental factors and other personal factors on PA level); therefore we could not examine the indirect effects of environment variables on PA and potential pathways between variables and PA.