Physical activity (PA) is important for health and a growing body of evidence points to the influence of neighborhood and community environments on population-level rates of activity and inactivity 123. Consequently, social ecological models are increasingly adopted in PA research and promotion that emphasize how numerous sectors and disciplines can contribute to improving active living behaviors 45. Within this new paradigm, parks and open space are acknowledged as key behavior settings 6 that can facilitate or discourage PA depending on both their availability and design 789. For example, in one study of travel diary data from more than 3000 children and teenagers in Atlanta, having at least one recreation or open space land use within 1 km from home was the built environment variable most consistently related to walking at least once over a two-day period and to walking greater than 0.5 miles per day 10. Another study found that the number of features within a park was more important than its size or distance from residents in predicting use of the park for PA, and that certain features were more strongly-related to park-based PA than others 11.

A key issue in the emerging PA and built environment literature concerns how best to measure environmental correlates of PA 312. Such debates include whether subjective or objective indicators are superior and the degree to which these different measurement modes correspond for the same objects being examined (e.g., nearest supermarket). Some recent studies have examined the level of agreement that exists when residents are asked to rate features of their neighborhoods versus when objective indicators such as observational audits or geographic information systems are used to characterize the same areas. These have generally reported low correspondence between subjective and objective indicators for most environmental features 1314151617.

Likewise, the few studies that have included perceived and actual measures of park proximity have reported only poor to fair levels of agreement 141516. For example, Macintyre et al. (2008) calculated a kappa value of .095, which is considered poor 18, when comparing participants' reports of living with a half-mile of a park with GIS measurements of the presence or absence of a park within a half-mile buffer. Examining the degree of correspondence between subjective and objective indicators of park availability or proximity is important because it has been argued that people cannot make use of neighborhood resources or destinations for PA if they are not aware of them 131619.

However, low levels of agreement may be attributable to the way parks have been treated in the majority of the literature to date. For example, almost all previous studies of park proximity, whether using subjective or objective measures, "have generally by default considered all parks and playgrounds to have the same elements and qualities, despite the awareness that they may differ substantially on these characteristics" 20. This is also true of the few papers that have directly compared perceived and objective proximity to parks within the same study 14151617. It has been noted that perceived distances and actual ability to predict distances may be influenced by the attractiveness of an end destination 21. Thus, it may be that residents are unaware of nearby parks which are smaller or contain fewer features, but that greater awareness and greater correspondence with objective measurements (and consequently, perhaps, greater use for PA) may exist for parks that are larger or possess certain unique characteristics. To date, this has not been empirically examined in the parks and PA or broader built environment literature. Likewise, certain demographic (e.g., age), neighborhood (e.g., safety), or other park-related variables may influence the extent to which residents are aware of the available parks in their neighborhoods 17.

Finally, among the relatively few studies that have employed comparable measures of both perceived and objective environmental features in examining how each is related to physical activity, neither objective nor subjective measures have consistently shown a stronger relationship with study participants' PA levels 22232425. With respect to parks specifically, one study found that self-reported greenness of one's neighborhood was significantly related to walking trips, but objective greenness was not 26. Another reported that neither subjective nor objective indicators of proximity to a park were related to participants' meeting PA recommendations 23. Mowen et al. 19 reported that perceived park proximity had significant direct and indirect relationships with reported park visitation, daily physical activity, and perceived health, whereas an objective measure of park proximity was related only to respondents' length of park stay. However, in spite of these few analyses, the majority of studies examining parks and PA have used subjective measures 27 and few others have included both objective and subjective indicators of park proximity in the same study to compare their relative importance in predicting PA. Better understanding whether (perceived) awareness, (objective) availability, or both is more important for predicting use of parks for PA may be valuable for knowing where to target efforts to promote PA through parks 1317.

Table 1 shows descriptive characteristics for the study sample. Participants varied in age and were grouped into two categories - 18 to 39 years (39.9%) or 40 to 88 years (58.9%). Over half were female (55.4%) and the majority of respondents were either married or living in a common law relationship (77.2%). Two-thirds of respondents (66.0%) possessed at least a college level education, just over one-quarter (28.4%) lived in a household that contained a child under 12 years, and almost one-third (31.9%) owned a membership to a fitness facility. With respect to body mass index, approximately half (48.1%) were classified as being a healthy weight, while the other half (49.3%) were either overweight or obese. Most of the respondents (78.6%) reported participating in at least some neighborhood-based physical activity, whereas a much smaller percentage (25.1%) engaged in at least some park-based PA during the course of the study week.

The primary purpose of this study was to examine the degree of correspondence between perceived and objectively-measured proximity to participants' closest park and how the level of correspondence varied according to several intrapersonal, neighborhood, and park-related variables. Achieving a match is important because people are unlikely to make use of PA resources of which they are unaware. Thus, it is important for park managers and other public health professionals to understand the public's awareness of facilities and services, what factors are associated with greater awareness, and how this familiarity might be enhanced.

In this study, we observed very poor agreement between perceived and objective proximity to parks, as indicated by a match for only 18% of participants and a kappa value of only 0.01. Other studies have reported similarly low levels of correspondence. For example, like our study, Kirtland et al. 17 reported a kappa value of 0.01 for survey and GIS measures for the presence of a park, playground, or sports field (grouped together) within 10 miles. Somewhat better, another study reported a kappa value of 0.39 ('fair' agreement) for a self-report measure of whether a park was within a 5-minute walk and whether one actually existed within 400 m of the respondent 14.

Other studies have more closely examined the direction or source of the mismatch. In Macintyre et al.'s 17 study of whether people reported and actually lived within a half-mile of a public green park, 62% (408/658) of respondents correctly predicted the presence or absence of a park within 0.5 miles - 355 (54%) said they lived within a half-mile of a park and they actually did, while 53 (8%) said they did not live near a park and GIS measurements confirmed this. However, for the remaining 250 people (38%) for which a match between self-reported and measured distance to a park was not observed, 51 (8%) believed there was no park within a half mile when there actually was, while 199 (30%) believed they lived within a half mile of a park but GIS measurements did not confirm their perceptions. Although that study achieved somewhat higher levels of agreement (62%), but a similarly poor kappa value (0.095), the source of disagreement was actually somewhat reversed compared to the present study. In Macintyre et al.'s study, a much greater percentage of those who incorrectly perceived the presence or absence of a nearby park were people who believed they lived within a half-mile of a park but really did not. In contrast, the present study reported a lack of awareness in that 79% of the sample (or 97% of the mismatched respondents) lived within 750 m of a park but stated they did not. Our results are similar to those of another study in which 76.5% of respondents overestimated the distance to their closest park (when given several categorical distance options), 15.3% correctly estimated the distance, and 8.2% underestimated the distance 16. Another study reported a kappa value of 0.15 in which 81% of respondents to a survey said they lived within 0.5 miles of a park, whereas only 62% actually lived with 0.4 miles of a park 26. Overall, the sources of disagreement have varied but past studies have generally reported poor correspondence between perceived and objective proximity to parks. These findings are consistent with conclusions from geography that people misestimate distances relative to their place of residence 33.

In this study, several variables were significant predictors of increased or reduced odds of achieving a match between perceived and objective proximity to parks. For example, with respect to socio-demographic characteristics, there was better correspondence for people who were younger, more educated, had children, and were a healthy weight. Further, a match was more likely among people who did not own a fitness facility membership and those who had engaged in at least-some park-based PA, with the latter finding being not surprising and perhaps suggesting a cyclical effect between park use and awareness. Similarly, a past study reported that more active people showed greater agreement for subjective and objective measures of proximity to recreation facilities, but not for many other environmental resources 15.

We also found that greater perceptions of neighborhood cohesion and lower perceptions of neighborhood aesthetics were associated with increased odds of achieving a match. Much research has suggested that constructs related to social capital have important implications for health 343536. In showing that perceptions of cohesion also increase awareness of neighborhood resources for PA, our findings add to the importance of urban planning that fosters improved bonds among residents and their communities. On the other hand, the finding about aesthetics is counterintuitive. It may be that persons living in aesthetically pleasing environments have less reason to seek out the environmentally-friendly confines of parks, but this and other hypotheses require exploration in future research. Finally, perceptions of neighborhood safety were unrelated to achieving a match. Past research has shown safety to be an important influence on PA behaviors 373839, but such perceptions may not influence simple awareness of resources.

Finally, the level of correspondence between perceived and objective proximity was a function of several park-related variables. The number of parks within 750 m and the number of features in the closest park increased the odds of achieving a match, while the size of and distance to the closest park were not significant. These results are consistent with past research about important influences on park-based PA 1130. As well, having a playground or wooded area in the closest park was positively related to achieving a match, while having a ball diamond or soccer field was negatively related to achieving a match. It is, as yet, unclear whether the size (e.g., a large wooded area) or prominence (e.g., a colorful playground) of park features, or both, are more important for improving park awareness. However, some research has suggested that the intensity of park-based PA varies by activity area 40 and that certain park features more strongly influence the likelihood of PA occurring in parks 11. Therefore, park awareness may be improved to the extent that the features in proximal parks are those that encourage use for PA or other active or non-active behaviors.

A secondary purpose of this study was to examine whether perceived or objective proximity to parks was more strongly related to a participant engaging in at least some neighborhood-based PA or at least some park-based PA. Objectively-measured proximity to a park was associated with neighborhood-based PA, but perceived proximity was not. It may be that having a park nearby (within 750 m in this case) facilitates an aesthetically and environmentally-pleasing setting that consciously or unconsciously encourages greater levels of activity in the vicinity of one's home. Another study found similar results in that the number of parks and the total area of parkland within 1 km of respondents' homes was related to greater levels of PA in the neighborhood 30. Future research should examine the direct and indirect mechanisms by which proximal parks facilitate PA in different settings.

However, neither perceived nor objective proximity to a park within 750 m were related to engaging in at least some park-based PA. A recent study also reported that objectively-measured distance to the closest park was unrelated to engaging in park-based PA 30. Likewise, with respect to perceptions, most studies have shown a general lack of awareness of nearby parks which would assumedly limit their use for PA by residents. Several studies have suggested that the content and design of parks may be more important than their proximity for encouraging PA 811. For example, one study of parents reported that only 49% frequented the closest park to their starting destination, and instead, the majority chose to travel a significant distance (in some cases over 4 km) to attend parks with certain amenities 41. Other research is starting to illuminate which features of parks are more conducive to promoting PA 11404243. Park planners should keep in mind that neither perceived nor objective proximity may be the most important factors to consider when designing parks to encourage PA.

However, when a match occurred between perceived and objective proximity to a park within 750 m, a significant positive relationship was observed with the likelihood of engaging in park-based PA. Consequently, complementary efforts should be undertaken to both locate parks appropriately and promote their availability and features to residents 19. A past study showed that even among people who had heard of several given parks, very few were knowledgeable about the specific features contained within them 44. The budgets of park planners for promotional materials are often limited, but the present findings suggest that investments to communicate the whereabouts of parks and their specific features may pay off in terms of improved use for PA.

There were several important limitations to this study. First, we did not know or take into account participants' length of residence. This may or may not be a key determinant of correspondence but it is likely that awareness of proximal resources for PA would improve the longer one lives in a particular area. Future studies should investigate the veracity of this notion and potentially control for temporal variables or exclude respondents who have not lived in the neighborhood for a minimum period of time. As well, because of the quantitative survey data collection method, we do not know what people were considering to be a park when estimating the distance to their closest resource. As noted, we had concerns about such definitional issues for estimating correspondence and thus elected to exclude nine fairly inconsequential parks from the present analyses. This did not appear to improve the kappa value observed in our analysis compared to past studies, but did provide greater confidence that the 45 parks that we retained from the GIS database better reflected what residents would actually perceive as a park. However, future studies should use qualitative or other methods to investigate how residents perceive and define parks and what factors limit or enhance their awareness of these neighborhood resources 17.

Additionally, when examining the association of perceived and objective park proximity with two different PA indicators, we collected only seven days worth of PA data. The detailed data collected via the log booklets were invaluable for determining locations of PA episodes, but also limited the time frame we could expect residents to participate in the study. Finally, we did not examine residents' psychological attachment to parks or the meanings they attribute to neighborhood open spaces. Although we found that several personal, neighborhood, and park variables were significant predictors of correspondence, it may be that people for whom parks are more significant may demonstrate better judgments about distance.

The findings of this study suggest that little correspondence exists between perceived and objectively-measured proximity to parks. Therefore, future researchers should be careful not to equate perceptions with actual distance. Just as importantly, given the lack of awareness about parks within 750 m, it is important to avoid assuming that actual distances are equivalent to perceptions 17 and that objective measures of proximity are necessarily superior. Despite these dismal correspondence findings, some encouraging results were observed in that participants who achieved a match between perceived and objective proximity to their closest park were more than one-and-a-half times more likely to engage in at least some park-based PA. Most past research has simply focused on perceived or objective proximity to parks without considering the inter-relationship between the two and their combined implications for active behaviors. Our finding suggests that both provision and promotion of parks are important for encouraging their use for PA. Further, the level of agreement was better for some sub-groups than others, suggesting that self-report measurements may be more accurate assessments of park proximity for certain people. Likewise, several park-related variables were related to increased or reduced odds of achieving a match between perceived and objective proximity. To our knowledge, this is the first study to examine correspondence according to the characteristics of nearby parks, but as research in this area accumulates, park planners would be wise to take note of the features that appear to improve awareness and use of neighborhood parks.

The authors declare that they have no competing interests.

Both authors participated in conceptualizing the paper, conducting analyses, and drafting and revising the manuscript. Both authors have read and approved the final manuscript.