A Participatory GIS-Based Method To Construct Output Areas For Neighbourhood Analysis And Social Planning

A Participatory GIS-Based Method To Construct Output Areas For Neighbourhood Analysis And Social Planning

(Version 8/21/06)

Nigel Trodd  and Wayne M Gearey Jr.

ABSTRACT: Strengthening participation in the planning process is considered to be a major challenge to better governance and by overcoming this challenge it is hoped that we, as a society, can make progress towards more sustainable development. It is widely believed that GIS has an important role to play, as recognised by the efforts being made to develop participatory and community-based GIS. It is perhaps less well documented what factors control the success of these initiatives. The aim of this paper is to explore how GIS can support a participatory approach to the definition of neighbourhoods and in so doing make a contribution in the early stages in developing evidence-based social policy. In developing a participatory GIS-based method the research identifies and investigates various factors, including the identification of relevant local knowledge for designing purposeful zones and GIS techniques that support participation. These factors are examined for a study of early child development and the targeting of pre-school services. The research then considers the portability of the method to similar studies in social epidemiology. It is shown that the approach can facilitate effective participation by a group of community advocates to delineate purposeful neighbourhoods and the use of census dissemination areas supports the reuseability of the method in other social studies conduted at the neighbourhood level.

INTRODUCTION

GIS in participatory social planning

In addressing the complexities of city planning it is important to remember that the physical form of the city sits alongside less tangible economic, social, environmental and cultural factors. Typical issues of public concern include access to public transport, congestion and air pollution, employment, crime, conservation, health and education. The ability of individual citizens and communities to share their understanding of the present and visions for the future is an important pre-requisite to informed planning and, through this, to building a consensus on complex issues such as sustainable development (Curwell and Hamilton 2003). It is disappointing then to find that participation in planning is often limited to consultation documents and public meetings and occurs at the later stages of the planning process. Skeffington (1969), DETR (1998) and others have argued that planning should involve stakeholders in the community putting forward ideas and comments earlier in the process to secure their ownership of and commitment to the planned developments. Early activities include prioritising planning goal(s), agreeing method(s) and defining a data model that is fit for the intended purpose. The relevance of the data model is defined by the variables to be measured and their (spatial) unit of measurement.

Carver (2003) has identified a ‘blossoming’ of interest in participatory-GIS and uncovered evidence that social science and GIS are ‘coming together to collaborate on participatory approaches’. Whilst this is promising he went on to question whether GIS data models are best suited to representing the more qualitative and perceptual effects of place ‘since this is more of a personal construct’. Social scientists might argue that in social planning we have more need for models that represent the geographies of communities than those of individual citizens. If we consider a neighbourhood to be the natural spatial expression of a community then “neighbourhoods provide a useful lens through which to examine wider social and political processes” (Statistics Canada 2005). In implementing a policy we apply governance to a geographical area or, as Crampton (2004) observed, “in order to govern, certain knowledge must be acquired about how things lie (including their spatial disposition)”. Furthermore, there is evidence to suggest that communities respond more positively when they can ‘see’ the consequences of their planning decisions at a local level (Carver 2003).

The aim of this paper is to explore how GIS can support a participatory approach to the definition of neighbourhoods and in so doing make a contribution in the early stages in developing evidence-based social policy.

Zone design and neighbourhood analysis

Without geographical structure, statistical data lacks context. It follows that, for data to be relevant, geographical reference must be provided. As a result a spatial unit is an intrinsic part of a survey with different standard units designed for different purposes. The data user has to map those units to relevant geographical areas. In social studies increasing interest has been shown in the neighbourhood as the standard spatial unit for small-area analysis. There is, however, a debate regarding the definition of a neighbourhood because it can be viewed as either a spatial or social phenomena and it evolves through societal processes operating over different timescales and at multiple scales. As a result their boundaries may be difficult to define (O’Campo 2003).

Automated approaches to zone design focus on geostatistical solutions which have the potential to become snared by the problems of the modifiable areal unit and can be perceived to lack purpose (Openshaw 1996). An alternative method to delineate neighbourhoods could be to capture local knowledge of social, economic, environmental and cultural patterns and employ an intelligence-led approach to create an agreed set of zones (Gearey et al. 2005). For example, a number of studies have aggregated census data into ‘purposeful’ units for contextual analysis e.g. Coulton et al. 2001; Diez Roux 2001; Ross et al. 2004. It can only help these and similar studies if the spatial data models, such as census output areas, are designed to increase the usability of the data (Martin et al. 2001). For this reason the 2001 Canadian census dissemination areas (DA) were engineered using criteria for intuitive ‘visible’ boundaries, small size, compact shape, temporal stability and 400 to 700 inhabitants (Statistics Canada 2005).

Enumeration areas lend themselves to choroplethic mapping. Whilst this technique has frequently been used to map population statistics several weaknesses have been identified, most notably the assumption that the population is homogeneously distributed within the geographical unit. This can limit its use for small-area analysis. An alternative to choropleth mapping are dasymetric techniques that use ancillary data to redistribute population data and correct for error, termed ‘ecological fallacy’, that may occur with choropleth mapping (Crampton 2004). The use of GIS opens up a range of new opportunities to secure interactive participation in the design process. Stakeholders are able to manipulate and query spatial data, select relevant layers to display, change the cartographic style and pan and/or zoom to areas of interest unaided or with the help of a facilitator.

It may be possible that communities could adapt a dasymetric approach, using GIS technology and guided by relevant ancillary data, to make a more informed decision regarding the delineation of neighbourhoods.

Early child development and the targeting of pre-school services

It is well recognized that early child development (ECD) is related to long-term health, well-being and general competence and adjustment in life. Longitudinal studies demonstrate that persistent trends in cognitive and behavioural development begin to emerge in children by early-to-middle childhood. Research on specific factors indicates that variations in ECD can be traced to early experiences within the family, peer group, school setting, community and cultural context (Human Early Learning Partnership 2006). Such findings have given rise to proposals to increase investment in social and economic programmes for pre-school children in Canada, the United Kingdom and the United States (Her Majesty’s Government 2003; National Research Council Board on Children 2000). A challenge is how best to target resources at vulnerable children.

Researchers at the Human Early Learning Partnership (HELP) use GIS technology to map neighbourhood level influences on individual pre-school children and relate those observations to the results of an assessment of the child’s ‘readiness-for-school’. Measures of child development are assessed using the Early Development Instrument (EDI).  In this way the Partnership has been able to identify communities at risk and present them in an Atlas of Child Development for the Canadian Province of British Columbia (Kershaw et al. 2005). In creating the Atlas and in developing strategic programmes for early childhood intervention there is a need to delineate comparable neighbourhoods at the local level.  Because units should be agreeable to their communities and suitable for the purpose of social area analysis and planning, input to the delineation process has been sought from multiple stakeholders who represent various aspects of the communities’ interests.

Based on the evidence of previous studies of participatory social planning and zone design and the requirements of the HELP it was postulated that a participatory approach to zone design for the small-area study of ECD and the targeting of pre-school services could be developed using GIS and census dissemination areas.  To assess this hypothesis, the research addressed three objectives:

1. Identify and prioritise local knowledge for designing zones for the study of early child development and the targeting of pre-school services.
2. Adapt a GIS-based dasymetric technique to support participation in zone design.
3. Assess the portability of the new technique.

METHOD

Case study: early child development in the Corporation of Delta, British Columbia

In order to investigate participatory approaches to zone design using census data, a case study was developed with the assistance of the Human Early Learning Partnership. The decision to locate the study in the Province of British Columbia was underpinned by the tradition of public participation in governance in Vancouver, as it was believed this might provide fertile ground for furthering knowledge of participation at an early stage in a social planning process. Furthermore, Statistics Canada had introduced dissemination areas in the 2001 Census units that, amongst other reasons, were intended to be suitable for defining areas of interest.

This study focused on School District 37, Delta, part of the Greater Vancouver region (Figure 1). Delta was selected on the basis that in the 2001 census its demography and ethnicity were similar to the average for British Columbia and over 1,100 children had taken the EDI test in 2003 (Table 1). Although fairly typical of communities in British Columbia it was noted that fewer children had been rated as ‘vulnerable’ on one or more EDI measures than the Provincial average (20.0% compared to 24.5%) and there was a well-organised Child and Youth Committee (CYC) who presented an opportunity to capture local knowledge.

Delta CYC takes the lead in planning and coordinating services for young children. Most of the individual members are from community advocate groups (Table 2). As a member of the CYC the HELP was interested in working with community advocates to delineate neighbourhoods that would enable the Partnership to analyse results of the EDI test given to pre-school children and the CYC to advise on the targeting of resources. In the case study, local knowledge (including the location of children assessed using the EDI) was captured and processed using GIS to draw neighbourhoods that were suitable for these purposes.

Table 1. Demographics for School District 37 (Delta) and the Province of British Columbia from the 2001 Canadian census.

Table 2. Member organisations of the Delta Child and Youth Committee.

The research was conducted in three exercises; an initial exercise to establish criteria that influence the perception of neighbourhoods; an exercise to investigate factors affecting a GIS-based dasymetric technique for zone design; and a final exercise that examined a set of outputs from one running of the technique and compared them to standard geographical units to see how portable the technique might be. The exercises were videoed to provide a record of the process by which results were achieved.

Exercise 1: Assessing criteria used to define neighbourhood

The first exercise was designed to test the principle that different stakeholders prioritize different criteria when defining neighbourhoods for different purposes. The exercise asked individuals to assess the importance of different criteria in defining a neighbourhood. Participants were asked to rate physical, socio-economic and geopolitical characteristics on a Likert scale from very weak to very strong. They were presented with two scenarios. First, each participant was asked to consider the importance of characteristics with reference to how they define the neighbourhood where they live; and second, they considered the importance of criteria to create units for analyzing ECD and the targeting of pre-school services.

Exercise 2: Mapping neighbourhoods using a modified dasymetric approach

The aim of the second exercise was to investigate two factors that might affect the performance of a GIS-based dasymetric technique for zone design. The first factor was the use of GIS and ancillary data to define boundaries; and the second factor was the effect of working individually or as a group.  The exercise required participants to complete two tasks. In the first task individual participants were provided with geospatial data for important criteria as identified in Exercise 1 and asked to use their local knowledge to construct neighbourhoods for analyzing ECD. In the second task the Committee was asked to work together and use its collective knowledge to define a set of neighbourhoods. The results could then be used to assess how robust the technique might be.

Exercise 3: Comparing neighbourhoods and census dissemination areas

The operational use of the technique was examined in the third exercise. Neighbourhoods defined by the Committee in Exercise 2 were compared to census DAs. This study was designed to identify conditions in which DA did not match the perceived community boundaries and so provide evidence on the portability of the technique for the analysis of ECD in other School Districts.

RESULTS

Exercise 1: Assessing criteria for defining neighbourhood

Eleven participants, representing all member organizations of Delta CYC except the Corporation of Delta, completed the survey.  Key results are presented in Table 3. The median was calculated because of the small sample size and more detailed analysis using the change in mean rating was made only for variables that saw a change in their median rating.

Table 3. Importance of criteria in defining personal and purposeful neighbourhoods.

The results show that neighbourhoods were defined, not unexpectedly, on the basis of numerous criteria.  Personal neighbourhoods were found to be the construct of physical and socio-economic characteristics. Neither geopolitical criteria had much influence. In contrast, there was evidence that some physical criteria were considered of less importance when defining neighbourhoods to analyse ECD. Churches, railways, rivers and streams were all rated ‘neutral’, the influence of rivers and streams being downgraded with the change in purpose.

Occupation, a socio-economic criterion, post code area and census dissemination area were considered more important for zoning the results of the EDI. The increased importance of the two geopolitical criteria was substantial. For statistical reasons the areas used to analyse ECD have to include at least 40 children who have taken the test and the community advocates strongly believed that these areas should also have approximately equal numbers of children.

One set of landmarks, elementary schools, were also considered more important when the purpose of the zoning was to assess ECD. This may be because of their association with the conduct of the EDI as it is administered to children in their first few weeks at elementary schools.

Exercise 2: Mapping neighbourhoods using a modified dasymetric approach

The individual participants used relevant ancillary data to draw neighbourhoods for the purpose of analyzing ECD. One participant argued that Delta was a single community and did not draw any neighbourhoods. The other participants used three different cartographies to delineate neighbourhoods – three used simple geometries e.g. ovals, rectangles; two used irregular polygons; and five used a combination of simple geometries and irregular polygons.

When the Committee members worked together they chose to delineate polygons using irregular polygons (Figure 2).

Exercise 3: Comparing neighbourhoods to census dissemination areas

GIS technology was used to compare the neighbourhoods defined by Delta CYC to standard Canadian census DA. The two sets of polygons were intersected and a measure of their correspondence was computed. The calculation assessed the percentage of a DA that fell within a single neighbourhood. If 99% or more of a DA fell within a neighbourhood it was considered to be a match for that neighbourhood (Figure 3).

The results found that only 9 (of 122) DA did not correspond to a neighbourhood. Five of these were associated with the area around Burns Hill and closer inspection revealed that some of these DA did not meet one or more of the design criteria set by Statistics Canada. For example, they were not compact.

In a separate exercise the neighbourhoods were endorsed by an expert in social epidemiology (Gearey et al. 2006) and the method was used to inform the definition of a geographical data model for the British Columbia Atlas of Child Development (Kershaw et al. 2005) (Figure 4).

ANALYSING THE OBJECTIVES

This research set out to address three objectives: (1) to identify and prioritise local knowledge for designing zones for the study of early child development and the targeting of pre-school services; (2) to adapt a GIS-based dasymetric technique to support participation in zone design; and (3) to assess the portability of the new technique. In analysing these objectives the research offered some insight into how a participatory GIS-based method could make a contribution in the early stages in developing evidence-based social policy.

Prioritising local knowledge in zone design

The results of this study have shown that individuals have their own perceptions of neighbourhood and that different criteria are prioritised depending on the purpose of zoning. This has been found in other studies (e.g. Coulton et al. 2001) and can be explained in terms of “personal construct theory”. Personal constructs, transparent patterns or templates that simplify the complexities of the real world, are formed during interactions with the physical world and from social interaction. In the use case the neighbourhood became a personal geographical construct that a participant used to make sense of his/her environment. Adding a specific purpose to the delineation of neighbourhoods was accompanied by a change in the priorities given to physical, socio-economic and geopolitical criteria. Defining neighbourhoods for the purpose of analysing ECD and targeting pre-school services saw an increase in the importance given to socio-economic and geopolitical criteria.

When individual participants applied their knowledge to delineate neighbourhoods they used a variety of cartographical styles. This was, perhaps, not surprising given the lack of guidance given to this part of the technique. Although the appearance of maps created by individuals differed there was considerable agreement in their core areas. This was reassuring as it was discovered later that although the participants worked in Delta only two of them were current residents in School District 37. These results suggest that identifying and prioritising criteria standardises the delineation process and increases consistency. Further research is necessary to investigate in more detail whether the differences between individuals’ maps are only cosmetic.

When working as a group the dialogue between participants enabled gaps in an individual’s knowledge to be filled. Neighbourhood boundaries were drawn more precisely on the Committee map. In addition, the video evidence suggested that social interactions taking place within the group were accompanied by ‘norming’ – both of the cartographical output and, more significantly, of an understanding of the purpose of the zoning. There was evidence that tapping into a socially cohesive community network gives the researcher a more informed way of identifying neighbourhoods used with GIS.

Portable GIS-based technique to support participation in zone design

The first two exercises showed that using a participatory GIS-based method provided an avenue for local knowledge, arising from social narratives, to be converted into geodata that was then used to successfully delineate “purposeful” neighbourhoods for contextual analysis. The research demonstrated how an intelligent partnership between geospatial technology and people can be created by letting people do what they do best, for example lateral thinking and making decisions using multiple criteria, and machines do things like repetitive and consistent pattern sifting.

This study showed that design criteria used to engineer census DAs have created a system of spatial units that could be used to delineate neighbourhoods. The level of agreement between DA and neighbourhood boundaries was greater than 90%. Further analysis found that had block census units been used then the match would have been 100%. Unfortunately, block census units cannot be used in all other studies because census data are not generally released at this level of spatial granularity.

Previous studies have shown that census units can act as proxies for neighbourhoods in small area analysis (e.g. Ross et al. 2004).  It was less clear, however, whether the neighbourhoods were ‘meaningful’ to residents (Ross et al. 2004), or whether residents looked at their neighbourhood in terms of census geography (Coulton et al. 2001). The first exercise in this study found that, for a given purpose, census units can have meaning and results from the other two exercises suggest that census DAs closely match the community group’s perception of their neighbourhood boundaries. Most areas in disagreement were found to involve census DAs that did not appear to meet one or more of the design specifications published by Statistics Canada e.g. visible boundaries, compact shape.

Zone design and evidence-based social policy

According to Carver (2003) the role of participatory GIS is to help minimise conflict and arrive at decisions that are acceptable to the majority of stakeholders through consensus building approaches based on awareness of the spatial implications of a decision. A consideration in developing a strategic approach to ECD is understanding and defining the levels of intervention that are available across communities (Kershaw et al. 2005). Interventions can be planned for an individual child (clinical intervention) or targeted at a subset of families and children. At the other extreme civil society interventions are designed to make communities ‘friendlier’ to families and more supportive in preparing children for schooling. In between universal programmes are available to everyone but not all families may be able to access them (Figure 5). On the evidence of this research a participatory GIS-based approach to constructing neighbourhoods could contribute to the early stages of planning universal and civil society interventions and in making targeted interventions.

This study joins a relatively short list of examples of real-life usage of GIS technology within participatory exercises. Carver (2003) speculated on reasons for the small number of examples. Contrary to his view there is no evidence from this study that well-informed community representatives were ‘not ready for it (PPGIS) yet’. The research has benefited, however, from longstanding and mutually beneficial relations between HELP researchers and the Delta community and access to GIS resources and expertise.

CONCLUSION

In this paper we have attempted to meet Krieger’s challenge to “document the validity of our GIS methods and provide conceptual justification for the geographic levels we choose to study, as well as the measures we employ” (Krieger 2003). The participatory GIS-based method has been shown to capture local knowledge and apply it through a modified dasymetric technique to zone design for small-area analysis. The method is made portable by the use of census DA to build a neighbourhood data model that can support the targeting of pre-school social services. By engaging key stakeholders at an early stage in the planning process there is hope that greater consensus can lead to more effective social policy.

ACKNOWLEDGEMENTS

The authors appreciate the time and effort contributed by the community representatives and officials of the Delta Child and Youth Committee, British Columbia, Canada, and they gratefully acknowledge the support and encouragement given by Professor Clyde Hertzman and colleagues at the Human Early Learning Partnership, Canada, and Professor Louise Wallace, Professor of Psychology and Health at Coventry University, UK. The research used copies of ArcGIS and Feature Manipulation Engine donated by Safe Software. This paper was originally presented at the Urban Data Management Symposium 2006 in Alborg, Denmark.

REFERENCES

Carver, S. (2003): The future of participatory approaches using geographic information: developing a research agenda for the 21st Century, URISA Journal, 15,  61-71.

Coulton, C., J. Korbin, T. Chan, and M. Su (2001): Mapping residents' perceptions of neighborhood boundaries: a methodological note, American Journal of Community Psychology, 29, 371-383.

Crampton, J. (2004): GIS and geographic governance: reconstructing the choropleth map, Cartographica, 39, 41-53.

Curwell, S. and A. Hamilton (2003): The Intelcity roadmap. Report of the EU Intelcity project IST-2001-37373.

Department of Environment, Transport and the Regions (1998): Guidance on enhancing public participation in local government, London: HMSO.

Diez Roux, A. (2001): Investigating neighborhood and area effects on health, American Journal of Public Health, 91, 1783-1789.

Gearey, W., N. Trodd and C. Hertzman (2005): PPGIS in zone design for community development, Proceedings of the 13th Annual Conference of GIS Research UK, Glasgow, 440-447.

Gearey, W., N. Trodd, C. Hertzman and L. Wallace (2006) Participatory zone design: a method to support the geographic targeting of neighbourhoods at risk for adverse health outcomes, Social Science and Medicine, submitted.
Her Majesty's Government (2003): Every Child Matters, London: HMSO.
Human Early Learning Partnership (2006): http://www.earlylearning.ubc.ca/ Last visited 5 April 2006.

Kershaw, P., L. Irwin, K. Trafford, and C. Hertzman (2005): The British Columbia Atlas of Child Development, Vancouver: Western Geographical Press.

Krieger, N. (2003): Place, space, and health: GIS and epidemiology, Epidemiology, 14, 384-385.

Martin, D., A. Nolan and M. Tranmer (2001): The application of zone design methodology to the 2001 UK Census, Environment and Planning A, 33, 1949-1962.

O'Campo, P. (2003): Invited commentary: advancing theory and methods for multilevel models of residential neighborhoods and health, American Journal of Epidemiology, 157, 9-13.

Openshaw, S. (1996): Developing GIS-relevant zone-based spatial analysis methods. In P. Longley, et al. (Eds.), Spatial analysis: modelling in a GIS environment, pp. 55-73. New York: John Wiley and Sons.

Ross, N., S. Tremblay and K. Graham (2004): Neighbourhood influences on health in Montreal, Canada, Social Science and Medicine, 59, 1485 – 1494.

Skeffington Report (1969): People and Planning, Report of the Committee on Public Participation in Planning, London: HMSO.

Statistics Canada (2005): A brief review of research on neighbourhoods in Canada, Ontario: Research and Analysis Division, Statistics Canada.
National Research Council Board on Children (2000): From neurons to neighbourhoods: the science of early childhood development, Washington: National Academy Press.

ABOUT THE AUTHORS

DR NIGEL TRODD Is a Principal Lecturer in GIS & Earth observation by remote sensing at Coventry  University, UK. Dr Trodd’s GIS research has focused on the design and evaluation of a learning systems approach to participatory GIS, geospatial data integration and interoperability and the development of geospatial technologies for urban reconstruction and development. Other research interests lie in the areas of land cover dynamics and the development of novel approaches to image processing. He has taught in England, Wales and Australia, researched in southern Africa, China and mainland Europe and is a Fellow of UNIGIS International, the world leaders in e-learning for GIS professionals.

WAYNE M. GEAREY JR. is a recent PhD graduate of Coventry University, UK and research associate of the University of British Columbia, Canada. He is educated in Political Science, Social Planning, Surveying and GIS and began his career working as an assistant to a Canadian Member of Parliament advising on matters of social policy. He has more than 15 years experience as a GIS consultant and analyst in local and regional government in Canada and Nigeria and for environmental studies in Greece.

CO-ORDINATES
Dr, Nigel Trodd
Institution : Coventry University
Address : Priory Street, Coventry
Postal Code : CV1 5FB
Country : UK
Telephone number : +44 (0)24 76 88 76 88
Fax number : 
E-mail address : n.trodd@coventry.ac.uk
Website : www.coventry.ac.uk

Wayne M. Gearey Jr.
Institution : Coventry University
Address : Priory Street, Coventry
Postal Code : CV1 5FB
Country : UK
Telephone number : +44 (0)24 76 88 76 88
Fax number : 
E-mail address : apy111@coventry.ac.uk
Website : www.coventry.ac.uk