This study examined associations between perceptions of the physical environment and LTPA among socioeconomically disadvantaged women possessing varying levels of individual and social factors known to be associated with physical activity. The findings suggest that a physical environment perceived to be supportive was associated with moderate levels of LTPA amongst women with psychosocial characteristics considered less beneficial for physical activity. This relationship was not evident among women with more favourable psychosocial characteristics, suggesting that perceiving a supportive physical environment may be particularly important for women with less favourable psychosocial characteristics.
Among women with low psychosocial scores, the perceived physical environment was associated with moderate amounts of LTPA (120 minutes/week) but not with any (versus none) or high LTPA (≥280 minutes/week). This could be due to the categorisation of LTPA and the individual, social and environmental factors, which may have reduced precision to detect associations, although descriptive analyses identified associations in the expected directions. Alternately, these women may undertake activities that can be performed in their local neighbourhood (such as walking), and as a result are more aware of the characteristics of their physical environment. More active women may perform different types of physical activity (e.g. organised sport, or walking, running or cycling longer distances) undertaken outside the neighbourhood, reducing their awareness of local neighbourhood characteristics. This is plausible as for women reporting 120–279 minutes/week LTPA, a greater proportion comprised walking (62%) compared with women reporting ≥280 minutes/week (53%). Another Australian study also found a greater proportion of participants walked for recreation within their neighbourhood than outside the neighbourhood, and that recreational walking within the neighbourhood was of greater duration than recreational walking outside the neighbourhood . This suggests that amongst women with low psychosocial scores, perceiving a supportive environment may be important for achieving levels of LTPA that approximate the recommended 150 minutes/week [28, 29]. Plausibly, women who have high psychosocial scores may be active irrespective of their environment, whereas for women with lower psychosocial scores, perceiving a supportive environment may be necessary to influence LTPA.
This study is limited by its cross-sectional design. It is plausible that active women perceive their environments as more supportive of activity or are more aware of their physical environment, or that women with more favourable psychosocial characteristics (e.g. higher physical activity self-efficacy) may over-optimistically report physical activity. Self-reported measures of LTPA and environments may limit findings. However, there was variability within psychosocial and environmental score categories, and associations were observed with well-established correlates (e.g. age, BMI, education, employment status, number of children)  in expected directions. While a reliable and valid measure of physical activity was used (the IPAQ-L) , the psychometric properties of specific domains of activity such as during leisure time, have not been assessed. Lack of specificity in the environmental measures in relation to LTPA might have reduced the ability to detect associations . For instance, the neighbourhood ‘physical activity environment’ refers mostly to walking activities, but the measure of LTPA is not limited to walking. However, walking is the most common leisure activity amongst Australian women  and more than 60% of participants reported walking for leisure. The large number of comparisons may be considered a limitation. However, by reducing the likelihood of false positives (a type I error) by adjusting for multiple comparisons, the likelihood of false negatives (a type II error) is increased, offering no real improvement .
While lack of access to a motor vehicle may have impacted the ability to access other more supportive environments further afield, only 6% of participants reported not having access to a motor vehicle (car ownership is high in Australia, with approximately 740 registered passenger vehicles per 1000 residents in Victoria ) access to a motor vehicle did not meet the definition of a confounder. We did not adjust for income, but previous analyses of data from this study found no association between income and LTPA , and hence income did not meet our definition of a confounder and was not considered as a potential covariate. It is plausible that there was residual confounding from the use of self-reported measures, unmeasured variables, or variables not included in our analyses.
This study was novel in its approach to understanding the complex relations between perceived individual, social and environmental factors and LTPA. While previous research has examined associations between individual, social and environmental factors and physical activity [8, 12, 13], no studies have examined the effect on physical activity when there are ‘mismatches’ in levels of individual, social and environmental characteristics. Other strengths include the large sample size which enabled assessment of a range of factors simultaneously while adjusting for known covariates and stratification by variables of interest, appropriate statistical adjustment for the multilevel data structure, and the use of a theoretical framework to guide selection of correlates. Our work focused on an under-studied population group at high risk for inactivity and chronic disease, and provides insights for targeting and development of interventions to promote LTPA among women living in socioeconomically disadvantaged areas.