Do residents’ perceptions of being well-placed and objective presence of local amenities match? A case study in West Central Scotland, UK

In recent years there has been growing interest in the significance of neighbourhood effects on health, and how particular features of the places in which people live may influence their health and well-being [1‐3]. Features of the physical environment may promote or impair health, either directly or indirectly, through the opportunities they provide for people to live healthy lives. Research has contributed to this topic by exploring the influence of accessibility to neighbourhood amenities and services [4‐6]. The accessibility of specific amenities can be theorised as contributing to health and well-being in many ways; easier access to shops, leisure facilities, health services etc. offers lifestyle choices and provides cost- and time-saving benefits which frees up resources for use elsewhere [7] and may offer opportunities for physical activities (e.g. walking to destinations within easy reach) [8].

Aspects of the neighbourhood environment have been measured in different ways. Previous literature has used self-report methods to survey residents’ perceptions of neighbourhood features by asking for example, whether parks or shops were within walking distance of respondents’ homes [4, 9]. Other studies have measured actual physical distance [5, 10‐12]. Objective measures may provide a more accurate picture of the neighbourhood physicality, but residents may not use local amenities or destinations if they are unaware of their presence, or if they do not consider them to be near or ‘local’. We therefore believe it is of interest to consider both subjective and objective measures of neighbourhood features; explore the extent to which they match and whether certain sub-groups show different relationships between the two (e.g. variation by sex, age, social class etc.). An understanding of the association between perceived and objective measures is essential, as if the former is used as a proxy for objective measures but the two bear only a weak relation then important associations may be overlooked, while others may be wrongly observed [13]. This may have implications for ‘Accessibility Planning’, that is assessment of whether people can reach key services (e.g. health care, education and food shops) within their local area; comprehensive planning should include subjective and objective measures of accessibility [14].

Most research to date has been based within regions of Australia and the US. Some of these papers reported a lack of agreement between perceived access to amenities and objective measures of access [6, 8, 15, 16], while others found significant agreement for specific features, such as, recreation facilities [17], supermarkets [18, 19], general food stores [16], schools [8, 20], and libraries, [8]. Few studies have examined agreement between perceived and objective measures of local area amenities within the UK [6].

In earlier work [6] we investigated the extent to which residents’ perceived assessment of living within half a mile of a local public green space corresponded to objectively measured distance, but found agreement was only 62% and the Kappa value (i.e. an inter-rater agreement statistic which evaluates the agreement between two classifications [21]) was poor at 0.095 (Kappa is significant if it reaches >0.20) [6]. In this paper we use data from a study based in West Central Scotland, UK, to, first, investigate whether there is correspondence between residents’ assessments of how well-placed they are for everyday amenities (food stores, primary and secondary schools, libraries, pharmacies and public recreation or sports facilities) and GIS-modelled accessibility measures. Secondly, we examine agreement by various socio-demographic sub-groupings as previous literature has shown variation in perceptions of neighbourhood features by characteristics such as sex and age [22], social class and area deprivation [13, 19, 23]. Finally, we investigate whether factors which may influence the respondents’ level of experience of their neighbourhood would in turn affect the correspondence between subjective and objective assessments of accessibility, such as car ownership [24], owning a dog [25], walking within the local area [4, 26], years lived at current home [13, 27], and regular use of local amenities.

As most previous studies have focused on only one particular type of amenity, e.g. physical activity facilities or food retail environments, and potential variation by individual characteristics has been overlooked, our study extends this work by including a number of different amenities and explores sub-group variation. Hence, our study makes an important contribution to discussions on how to effectively measure and interpret residents’ access to local amenities.

For this paper, we used data from one aspect of a larger study: the ‘Transport, Housing and Well-being Study’ (THAW 2010). THAW 2010 was based on THAW 1997, a study designed to examine three objectives, firstly, the statistical associations between long term morbidity and mental health and well-being on the one hand, and housing tenure and car ownership on the other (while controlling for socio-demographic and psychological characteristics); secondly, the role of housing quality, residential environment, and use of cars, in influencing illness and psychological health; and thirdly the meaning of housing tenure and car ownership in people’s daily lives [28‐34]. THAW 2010 draws on respondents from similar geographical areas and uses a similar questionnaire to the previous study; for THAW 2010 a postal questionnaire, with three reminders, achieved a response rate of 38% (2092 completed questionnaires), from a random stratified sample of 5521 adults from the electoral roll in local authority areas in West Central Scotland. Survey respondents’ ages ranged from 17 to 95 years old. The survey included questions on the respondents’ mental and physical health and well-being, lifestyle, housing, neighbourhood, transport, employment, and finance. Compared to the West Central Scotland population, our study sample characteristics were broadly similar for sex and for age; 56% were female, and 65% were of working age (18 to 60 years old), compared to 52% and 62% respectively within West Central Scotland [35]. Within our sample, 85% of respondents had access to at least one car or van, while within the 2010 Scottish Household Survey, within West Central Scotland, 70% had access to a car (does not include van access) [36]. The socio-demographic characteristics of THAW 2010 were comparable to the previous THAW Study; e.g. respondents’ own social class was similar in THAW 1997 and THAW 2010 (65% and 70% in the non-manual social class groups, respectively). THAW 2010 was approved by the Ethics Committee of the Faculty of Law, Business and Social Sciences at the University of Glasgow.

In this paper we focus on six amenities: respondents were asked ‘How well-placed do you think your home is for’- ‘general food stores’, ‘primary schools’, ‘secondary schools’, ‘libraries’, ‘chemists or pharmacies’, ‘public recreation or sports facilities’. Although we did not define the term well-placed on the survey form, we expected that our respondents would interpret this term in accord with the standard definition as given in the Cambridge Dictionary as being ‘in a very convenient position’ [37]. We did not include the full set of amenities asked in our survey as some items referred to aspects of the physical environment which were difficult to measure independently e.g. ‘safe play areas’, or ‘pleasant green spaces’, or referred to resources that were fewer and further afield, e.g. ‘hospitals with a casualty department’. Respondents were asked to choose from one of four options for each resource: ‘very well-placed’, ‘fairly well-placed’, ‘not very well-placed’, or ‘not at all well-placed’. In previous research access to local amenities has been garnered using similar questions regarding how well-placed respondents’ homes are for, e.g., leisure facilities or health services [38‐40]. In this study, for the purpose of analysis, and to increase the size of comparative groups, the responses were re-grouped into two categories ‘very well-placed/fairly well-placed’, and ‘not very well-placed/not at all well-placed’. For comparison, we also grouped responses into ‘very well-placed’ and ‘fairly well-placed/not very well-placed/not at all well-placed’.

Various on-line and Ordnance Survey [41] resources were used to extract the addresses of the six amenities within buffers around respondents’ postal code locations; schools’ data was obtained from the Scottish Schools Online website [42], library details from the relevant Local Authorities websites, pharmacy details from the relevant Health Board websites, and general food stores’ and public recreation/sports facilities’ data were collected from Ordnance Survey Points of Interest data (2010) [43]. Ordnance Survey Maps (including addresses and roads, paths etc.) [41] were obtained and ArcMap Version 9.3.1 was used to geocode the location of every respondent by the centroid of their unit postcode (unit postal codes are the smallest level of postal geography in the UK and typically contain around 15 address points), and the six amenities. ArcMap was used to create both Euclidean and network buffers around respondents’ home postal codes. A Euclidean buffer is a straight-line circular radius around a point, while a network buffer includes a defined distance along the pedestrian road network (i.e. roads and paths) in all possible directions away from a point, and the routes’ end points are joined to form an enclosed area. ArcMap was used to create straight-line buffers of 800 meters (m) and 1000 m around (the centroid of) each respondent postal code; the presence (or absence) of (at least one of) each of the six types of amenities within each buffer was then recorded. For network buffers, the presence of the amenities within 800 m, 1000 m, and (due to some amenities being further away on average than others) 1200 m were determined. At an average walking speed of 5 km per hour [44], 800 m relates to (approximately) a 9–10 minute walk, 1000 m a 12 minute walk, and 1200 m a 14–15 minute walk. Similar buffer sizes have been used previously [11, 17, 45‐48].

SPSS Version 15.0 was used to investigate whether the objective presence of an amenity within a specified boundary was perceived by a respondent as being well-placed for that amenity, by cross-tabulating the objective and subjective variables (for each amenity) and calculating Kappa statistics. Kappa measures inter-rater agreement for categorical items, accounting for agreement occurring by chance (i.e. Kappa < 0.20 = poor agreement, 0.21-0.40 = fair, >0.40 = moderate to very good) [21]. Analysis was undertaken for all respondents together and also separately by sex; age tertile (17–45 years, 46–60 years, over 60 years); head of household Social Class (UK Registrar General’s Classification of Occupations [49]); and area deprivation (using the Scottish Index of Multiple Deprivation (SIMD) Income sub-domain score (based on numbers of claimants for a range of welfare benefits [50])). A look-up table was used to link the respondents’ postal codes to data zones, and for each data zone we obtained the SIMD Income score (SIMD 2009 scores divided into quintiles (Q1 = least deprived, Q5 = most deprived)). We did not undertake analysis separately by Urban/Rural scores as all respondents lived within accessible areas (93% within urban areas/accessible towns, 7% within accessible rural). We did not explore relationships between perceived and/or objective measures of access to amenities and physical activity as physical activity was not comprehensively measured within the study.

Some variables were included to measure respondents’ likely degree of experience of their local area. Thus, analysis was also undertaken by household car ownership; having a dog in the household; and whether the respondent took a walk of two miles or more in the previous year (and whether this was in or outside their local area), and years lived at current home in three groups (less than two years (<2), two to ten years, and over ten years (>10)). We included these variables to explore whether those who we assumed spent more time within their local area (i.e. non-car owners, dog owners, those who walked within the area, and those who had lived in their home for a longer period) would display better agreement.

Additionally we explored if respondents made use of supermarkets, sports facilities, and if they took children to school, and the form of transport they usually used for each. We do not know if this travel was to local amenities or those further away (data on travel to general food stores, libraries and pharmacies not gathered). We compared agreement between those who made use of the three amenities, and those who did not, and for the former we compared those who walked to the amenities and those who used other means of transport (i.e. car, taxi, public transport). Within the THAW Survey 1760 respondents had no missing values for any of the variables to be included in the analysis (with the exception of an additional eight missing responses for where respondents walked (inside and/or outside the local area), and two missing responses for time lived at current home).

Our results showed that residents’ perceptions of being well-placed for everyday amenities, such as food stores, schools, libraries and sports/recreation facilities, was poorly matched with geographically modelled access (or simply put, respondents were more likely to perceive an amenity as closer or farther than it physically was); the only exception was for pharmacies where agreement was fair (for 1000 m straight-line buffers).

Results varied according to the categorisation of well-placed, the resource examined and the objective distance employed. We found variation between the 800 m and 1000 m buffers, with the latter more likely to display fair agreement; and observed variation between the straight-line buffer and network buffer analyses, which may be due to the latter covering smaller geographical areas. However, it is conceivable that when respondents considered well-placed they were not considering specific road/path network journeys [6]. When analysing data using responses grouped into ‘very well-placed’ versus ‘fairly well-placed/not very well-placed/not at all well-placed’, results differed; a greater number of sub-groups displayed fair kappa values, for libraries and pharmacies in particular, and more so within the 800 m buffer than the 1000 m buffer (as seen in the sensitivity analysis).

Of the respondents who displayed poor agreement for food stores, primary schools and sports/recreation facilities, there were similar proportions of those who considered their homes well-placed but yet had no amenities within 1000 m, and of those who did not consider themselves well-placed but had an amenity within 1000 m. In contrast, for secondary schools, libraries and pharmacies, of those with a mismatch, the majority thought they were well-placed but had no amenities nearby; resulting perhaps from these amenities displaying the lowest representation across the buffers, and lowest overall density.

It is difficult to theorise why there would be fair agreement between perceptions of being well-placed for a pharmacy and objective measures, as pharmacies were not more abundant, and the frequency of trips to other local amenities, such as food stores, in general, may be higher [51]. It may be that pharmacies are more visible within the local community because of the important part pharmacists play in primary health care; pharmacies are an accessible health resource for everyone as they often have longer opening hours than GP surgeries, appointments are not needed, and they provide a range of services, e.g. dispensing of drugs, health and lifestyle advice, signposting to other services, support for self-care, medicine use reviews, health checks, access to emergency contraception, and addiction services [52]. It is also the case that the quantity of drugs prescribed in Scotland has risen by a third in a recent years (2004–11), so that the use of pharmacies is increasing; indeed, it is reported that 60% of the population of Scotland receive at least one prescription in any three month period [53].

Prior studies have compared objective and subjective neighbourhood features by socio-demographic characteristics, with varying results. Research in Australia found that males were more likely to show mismatch between perceived and objective access to supermarkets, but there were no significant sex or age differences for access to sports fields, or libraries [22]. In our sample, males showed better match for libraries and pharmacies than females. A study within neighbourhoods in Glasgow, UK, showed that females were much more likely to do food shopping in the local area [54], but in our study females were not more likely than males to exhibit better match for food stores. In the present study, the youngest age group was more likely to display mismatch, while older groups showed fair agreement for certain amenities; over 90% of the over 60 s were retired so may spend more time at/around home, and thus have better knowledge of local amenities.

In our study there was no clear pattern of agreement by Social Class or income but non-manual classes and those within more affluent areas showed poor agreement for all amenities. Previous research undertaken in West Central Scotland, found that those with lower incomes were more likely to shop for basic food items locally [55] and may therefore be more likely to use other local facilities. Furthermore, in the present study, non-manual classes and Q1 (the least deprived areas) had the greatest car access so would be able to go further afield for activities, such as food shopping. Indeed those without cars did show fair agreement for schools, libraries and pharmacies while car owners, and those who in general used cars to travel to supermarkets, sports facilities and schools, showed mismatch for all amenities, perhaps echoing findings from other studies that perceptions of travel time may be influenced by car ownership [24, 56].

Those respondents who had a dog in their household showed better agreement for primary schools and for pharmacies than those without a dog (at 1000 m). Owning a dog may encourage people to go for more walks in their neighbourhood [25]. Among those who had walked at least two miles in the local area, agreement was fair for primary schools and pharmacies, while those who tended to take walks further afield showed poor agreement for all amenities, this is in line with previous studies showing that people who engaged in leisure-time walking within their own neighbourhood were more aware of neighbourhood features [4, 26]. We found that those who had lived in their current home for less than 2 years showed better agreement for four of the six amenities studied; this contrasts to previous studies which found that newcomers took a longer time to orientate themselves with neighbourhood features [13, 27].

Overall, in this study, there was correspondence between respondents’ perceptions and the GIS measures for some but not all the amenities. The lack of agreement here, and similarly in other literature [6, 15, 16], could reflect that people generally inaccurately estimate how far or near a destination is [17, 57]. Previous work has shown that various factors could influence ‘cognised’ distance to environmental built attributes; e.g., a route with more intersections and barriers (i.e. buildings and trees) could lead to overestimation of recalled distance [58, 59], while an increase in physical effort, e.g. going uphill, could also lead to overestimation [60]. A respondent may appear to be objectively well-placed but access may be limited by a route which is deemed unsafe or undesirable leading to overestimation of perceived distance [23], while subjective distance to a resource could be underestimated if the respondent equates being well-placed with attractiveness of the destination [24, 61].

Respondents’ self-reports of being well-placed could be influenced by the fact they do not use local facilities regularly or at all. Respondents may not rely on their home location for amenities but use other non home locations such as areas around work, study, child’s school or transport hub [62]. For some people the area around home may be of little relevance to daily life, and this could lead to inaccurate perceptions of proximity [24].

Our study displayed a number of limitations. Our analysis is based on a general question, we do not know the distance respondents consider well-placed, and do not know whether the respondent regarded the question as relating to walking or driving. Nonetheless, we do not believe that using a more specific question would have provided more accurate responses, such as “is your home a 10 minute walk from a library?”, as walking could be affected by walking speed, mobility etc. Moreover, we do not know if respondents’ answers related to ‘spatial’ and/or ‘non-spatial’ factors e.g. amenity prices, quality of service, range of goods on sale, or opening hours [38]. We found variation in results according to how responses were categorised (i.e. combining ‘very well-placed’ and ‘fairly well-placed’ or including the latter only) which emphasises the difficulty of associating respondents’ ideas of well-placed with objective distance. We recognise that being well-placed does not necessarily equate with neighbourhood satisfaction [63], and that residential proximity may not be imperative in influencing usage of local amenities; the quality of a resource may be of greater importance but we do not have this data, nor do we possess data on which amenities respondents use. People may not use adjacent amenities to their home as they may be unsuitable, for example a study in Glasgow found that while participants agreed that objective access to green space was important, it was not the only factor which influenced usage. They discussed personal reasons for not using green space, e.g. being fearful of potential anti-social behaviour from young people present within a park; or simply not wanting to associate with others from the local community [64]. These barriers could influence the usage of other local resources. We did not explore links between perceptions and objective measures of the built environment and actual physical activity behaviours, e.g., utilitarian walking, as we do not have this data.

Our approach comparing objective and subjective measures of accessibility raises questions about the role of perceived convenience in local amenity use (the latter having potential benefits, for example, in terms of health behaviours and community vitality). Past research has shown that frequency of amenity patronage is influenced by perceived distance, and that choice of amenity/outlet is influenced by perceived locational convenience [65]. However, other research has drawn a distinction between different types of services, with locational convenience being more important for customer loyalty (when satisfaction is low) in the case of standardised, less personal services, and less influential upon customer choice in the case of less standardised, more personal services [66]. Our findings of a greater correspondence between perceived and objective convenience in the case of pharmacies, schools and libraries, therefore raises questions as to whether this is reflected in higher usage of these amenities and/or is reflective of a view of these amenities as being of a more or less personalised nature. These are two issues to be pursued in further research.

The findings also have implications for debates around ‘Accessibility Planning’ and policy formation. Accessibility Planning assesses whether people are able to reach the key services they need/are important to them (i.e. health care, education and food shops), and contends that comprehensive planning should include considerations of not only objectively modelled distance to key services, but also residents concerns about accessibility [14] e.g. how well placed do they feel. In this way barriers to access can be investigated [38] whether these relate to a lack of physical access to services, or residents simply not knowing what is available. Our study also has implications for the development of public health interventions which aim to increase active travel (walking, cycling) such as providing walking maps to increase awareness of proximity to community resources [67].

Despite its limitations we believe a major strength of this study is the contribution to the limited UK literature on comparisons of subjective and objective access to neighbourhood amenities, and inclusion of a variety of amenities covering health, education and leisure. As in previous studies [4, 13, 19, 22, 23, 26] our study highlights the importance of considering that agreement between perceptions and reality of proximity to amenities may vary when comparing certain population subgroups and socio-demographic characteristics.