We used mental mapping and GIS to explore congruence between peoples' perception of their neighbourhood area and definitions typically used in studies examining the association between environmental variables and physical activity.
Our results highlight important discrepancies depending on the neighbourhood definition used, with participants generally conceptualising a smaller neighbourhood area than those typically used in questionnaires. Perceived neighbourhood areas tended to be considerably smaller than those often used in physical activity research, from 1 km (or 0.5 mile) Euclidean buffer areas [23, 24] to those of 1 mile or more [15, 25–27]. Colabianchi reported that a 0.75 mile buffer was appropriate to define the walking neighbourhood within "easy walking distance" for older female adolescents . Our sample were adults, the majority of whom had access to a car for personal use and whose low levels of transport-related activity had been previously demonstrated . It is, therefore, reasonable that the average walkable neighbourhood area in our sample would be smaller than 0.75 miles (as the data infer) and closer to estimated '5-minute walking' distances used elsewhere (e.g., 0.25 miles , 400 m ). In this sense, it would appear that operational definitions of neighbourhood need to be smaller than those typically used.
On the other hand, analysis of the destinations showed that as a result of this size discrepancy, 42% of all walking destinations fell outside of areas that participants perceived as 'their neighbourhood'. It could, therefore, be argued that use of the larger 1 mile Euclidean buffer, which captured 96% of destinations, is acceptable despite its lack of congruence with the perceived area.
Although important for developing tools to measure environmental perceptions, the absolute size of the area to apply might not be the important issue. There was great variety in the size of individuals' perceived neighbourhoods, ranging from single streets, to areas including the local town centre and surround. This variation is likely to pose a greater problem. Regardless of size, imposing a simple uniform definition in studies of environmental perceptions should promote comparability and help to standardise the areal boundaries for what is a very subjective process (i.e. judging the presence, proximity or quality of characteristics within a given area). Yet there are lessons from the sociology literature that have yet to translate into environmental physical activity research . Mental mapping exercises by Chaskin  have shown that neighbourhood could be defined as a social unit, a spatial unit, or a network of relationships, associations and patterns of use, and that this has implications for size. For example, those defining their neighbourhood in terms of social relationships are more likely to describe smaller units, than those thinking of institutions and other frequently travelled destinations. Moreover, while individuals might stress one dimension over another, the area is rarely the result of a single dimension. Moudon and colleagues  who explored some of these concepts within the context of walking behaviour, stated that, 'Neighbourhood evokes socio-physical homogeneity, a shared sense of place, connection, and access. It has multiple cognitive, economic, geographic, behavioural, cultural, and temporal dimensions' (p.S102). This multi-factorial nature of defining your own neighbourhood could explain the marked variation in size and shape observed in the present study and reported elsewhere . It highlights the complexity of the neighbourhood concept and the challenge of measurement.
We were not surprised that the most popular local walking destinations were shopping/retail destinations. However, the importance of family and friends as a destination reported in 36% of participants suggests that it should be a feature of neighbourhood environment-walking/physical activity surveys. To date, it has been largely ignored, with few exceptions [31, 32]. Half of participants reported walking to green space, compared with just 5% who walked to physical activity facilities. The importance of informal recreation and access to quality green space has been reported in previous UK studies [7, 33–35] and appeared to be confirmed by the data presented.
Our findings have identified a number of issues that warrant further consideration by researchers. We need to better understand what people are thinking when we ask them questions about their neighbourhood environment. Despite offering standard definitions, the multitude of potentially influential social and cultural factors, both individual and areal, in addition to the context of the question (e.g. whether asking about the presence of trees on streets or pedestrianised areas), clearly results in wide inter-individual variation in the size of neighbourhood area. When developing surveillance tools, it will not be possible to take all of these factors into account, but further work to reach manageable data collection processes that improve on current practice in physical activity research is certainly warranted.
At least in relation to transportation walking, we have highlighted an opportunity to further explore walking destinations and the perceptions of the environment en route as a potential alternative or adjunct to 'neighbourhood' in physical activity and walking studies. The approach piloted here used maps with a level of detail that appeared fit for purpose and manageable from both researcher and participant perspectives. But this was to record the location of destinations only. To gather information on each of the routes (and there could be several for each destination, and many destinations) would represent a hugely time consuming and detailed process, greatly increasing participant burden. Therefore, the logistics of comprehensively capturing this information would be prohibitive for monitoring and surveillance, but there is scope for further work to turn this concept into a manageable and simplified, but valid data collection process.
The findings also make a case for similar work on a larger scale, including more qualitative evaluation, and for referring to other disciplines where such issues have been researched in more depth. By using lessons learned from sociological investigations, and using relevant approaches such as mental mapping, GIS, Global Positioning Systems (GPS) and cognitive interviewing, our understanding of this important area can be improved. Our study sample was too small to be able to identify any patterns of conceptual definition or neighbourhood destinations, which we would recommend for future investigation. We did observe much confusion and different abilities amongst our sample to identify their home location and neighbourhood on their maps, a further limitation of this approach. However, consistent with the views of others , we feel the feasibility and novel data produced, warrant pursuing and refining this approach in an effort to reduce the potential misclassification of local walking neighbourhoods.