GIS and Evidence-Based Policy Making - Chapter 15 (end) docx

A range of area-based policy initiatives designed to promote convergence of social conditions between communities have been developed, e.g., Health Action Zones, Education Action Zones,

Section II Engaging with the Public

15

Engaging Citizens: The Bradford

Community Statistics Project

Derek Reeve, Erik Thomasson, Steve Scott,

and Ludi Simpson

CONTENTS

15.1 Introduction 329

15.2 Policy Context 330

15.3 Project Context 332

15.4 Technical Features 334

15.4.1 Boundary-Free Small-Area Estimates 334

15.4.2 User-Defined Target Areas 337

15.4.3 Information Dissemination Rather Than Analysis 338

15.5 Data Issues 339

15.6 Community Use 341

15.7 Organizational Context 343

15.8 Conclusions 344

References 345

Engaging the active participation of citizens in the processes of civic govern-ance has been a laudable, if largely unrealized goal, of local governments for decades Recently, this goal has been much reemphasized in the United King-dom Almost every recent U.K government initiative places a clear onus on local governments to collaborate with their communities Local authorities must now establish community strategies and form local strategic partnerships

to reflect community interests In the main, however, the public have been

steadfastly disinterested in participation exercises Participation in local elec-tions and attendance at local meetings are usually depressingly low

The Internet has recently been seized upon as a new vehicle with which

to reengage the public Visionaries look forward to a future in which Internet-based systems will be used to involve citizens in developments in their locality, encouraging citizens to interact directly with professionals and policy makers

in local decision-making processes Because many of the issues that affect local government are land or property based, WebGIS are seen as having a major role to play within this movement towards Internet participation There is already a significant literature describing experiments in public participation GIS (PPGIS) (Craig et al., 1998; ESF-NSF, 2001; Laurini, 2001)

Achi eving the future, howeve r, is always more difficu lt than the visio nar-ies, and vendors, sugges t The rhetor ic surrou nding PPGIS has raced far ahead of re ality as rep resented by presen t PPGIS We bGIS package s pro vide the te chnologie s by which local ag encies might deliver spat ial informati on into the homes of citi zens but we still are at the very beginn ing of the lear ning curve of unde rstandi ng how to desi gn systems based on these techn ologies effici ently to engage the publi c’s intere st We need to unde rstand wh at inform ation shou ld be pr esented , how that inform ation is mo st effectivel y pre sented, and what is require d fo r the publ ic to be able to use the inform a-tion It is doubtf ul if any pre sent PPG IS could yet claim to have bec ome a major channel for participation between citizens and policy makers We are still at the stage of seeing what works, of experiments and projects

This paper contributes to the continuing PPGIS debate by detailing the PPG IS built for the Bradfor d Comm unity Statist ics Pro ject (BCSP ; www

purposeful PPGIS, the lessons from which should be of interest to both researchers and other local governments Some PPGIS sites appear primar-ily to be designed to disseminate prepared mapped-based information to residents and, with such sites, the manner in which the data are presented remains largely controlled by the sites’ owners The primary purpose of the Maps and Stats system, however, is to put into the hands of residents the datasets and online tools necessary to allow them, independently, to research conditions within their communities Our site invites users

to actively engage with data, rather than passively to receive them

A further distinctive feature of the BCSP is that the Maps and Stats PPGIS has been developed as one element within a broader project to build the capacity of local communities to understand and critically appraise the statistical bases upon which decisions about their localities are being made

The U.K government has rediscovered the policy significance of urban social geography There is presently very great concern about the spatial

dimension of social exclusion, the belief being that some localities are effectively excluded from the standards of well-being which are considered the norm in the rest of society Furthermore, there is a determination that such spatial inequalities will be reduced: ‘‘Within 10 to 20 years, no one should be disadvantaged by where they live’’ (Neighbourhood Renewal Unit, 2001)

A range of area-based policy initiatives designed to promote convergence

of social conditions between communities have been developed, e.g., Health Action Zones, Education Action Zones, Sports Action Zones, Excellence in Cities Action Zones, Sure Start, Anticrime local partnerships, the New Deal for Communities, and the Neighbourhood Renewal Fund PACT areas In addition to U.K government initiatives, quasi-government funding bodies, including the National Lottery Board, target significant resources into urban areas via area-based bidding processes The EU also uses area-based statis-tics to direct very large structural regeneration funds to selected areas Although each of these policies differs in detail, there are recurrent themes presented below:

. Territorial: There is an emphasis upon drawing boundaries Such territorial delineation is seen as a means of most effectively allo-cating limited resources There is, however, a clear equity issue with area-based policies: Communities within designated areas will qualify for assistance under particular initiatives, those out-side will not The onus is upon policy makers to ensure that boundaries are drawn appropriately

. Evidence-based: The mechanism for justifying the delineation of boundaries depends heavily upon statistical profiling To qualify for assistance, areas have to display specified characteristics Once designated, statistics are used to measure the progress of areas towards target norms A clear emphasis within recent policy making has been upon evidence-based approaches

Actually, there is nothing particularly new about this approach to urban policy making In the United Kingdom, there has been a tradition of small-area policy initiatives During the 1970s for example, there were housing action areas (HAAs) and general improvement areas (GIAs), the purposes

of which were to identify small pockets of need within cities and then to channel resources into the areas affected There is also, of course, a long tradition of spatially-based multiple deprivation and territorial social indi-cator studies, which have used statistical techniques to identify communi-ties in need (Knox, 1975; NCRNRD, 1998; Senior, 2002)

More novel strands within current small-area policy making might be: . Multiagency: There is an expectation that local providers of ser-vices to communities—local and central government departments,

health and police authorities—will coordinate their initiatives towards localities Cross-cutting initiatives are in vogue and the joined-up approach to government is being road-tested within the local policy-making arena

. Community involvement: A key feature of current local policy making

is a concern that communities should be active partners in the policies that are developed for their areas There is concern to reduce the alienation from governmental processes that are a characteristic

of socially deprived areas Within current policy initiatives, there-fore, there is an emphasis upon reengaging with local communities, although in practice community and voluntary groups may well retain a degree of cynicism about being the beneficiaries of another round of top–down policy initiatives

To realize such policy initiatives local governments and other local agen-cies are being required to reconsider the ways local decision-making is conducted On a pragmatic level, there is a great need to ensure that good quality small-area statistics are available to allow the characteristics of localities to be probed At a national level, the need for improved small-area statistics is emphasized within the National Strategy for Neighbour-hood Renewal initiative: ‘‘Better information needs to be available for all involved in strategy development, service design, and delivery at the local level This should make it more likely that problems are diagnosed and effective answers produced It also fits well with the need to involve local people more in playing their part and holding public services to account’’ (Social Exclusion Unit, 2000, p 8) And at the local authority level, there is an onus upon local agencies to share their datasets and to integrate their policy making more fully than has previously been the case

There is also a clear expectation placed upon local government to reener-gize its methods of public engagement Rather than going through the rites

of public participation, there is now an expectation that communities must

be genuinely active partners in formulating the policies that affect them

Against this national background, the BCSP is an attempt to enhance the capacity of community groups to participate more fully, and more equally,

in the local policy debates that affect their communities and, specifically, to help residents to understand the statistical manipulations involved in local area policy making and grant allocation procedures The Maps and Stats PPGIS is also seen as providing a platform that will facilitate the efficient integration and dissemination of previously disparate datasets

The BCSP is a joint initiative between the Research and Consultation Service of the City of Bradford Metropolitan District Council (CBMDC)

and the Bradford Resource Centre (BRC), the BRC being a not-for-profit organization that provides a focal point for community groups within the district The project was funded by a grant from the European Regional Development Fund (ERDF), with matching funds provided by the CBMDC

In its bid for funding, the BCSP provided a succinct statement of its aims: (a) Increase the capacity of community groups to effectively use stat-istical information sources

(b) Make accessible via the Internet local statistics for community groups’ own areas

(c) Support the voluntary sector in making a case for statistics which are more appropriate to their needs

To achieve objective (a) a team of workers was established within the BRC with a remit to liaise with local community groups and to foster among such groups confidence that they can make effective use of statistical data for their own purposes The BRC team provided informal ‘‘drop-in statistical surgeries’’ focused on helping community activists develop skills in the critical use of statistical sources and also developed a one-day ‘‘Strength

in Numbers’’ course to explore the issues in a deeper and more structured way Ongoing support has also been provided to those undertaking com-munity research Objective (c) sprang from a concern among comcom-munity researchers that official statistics often fail adequately to reflect the concerns

of community and voluntary organizations and that the community should have a role not only in interpreting existing statistics but also in influencing how and which statistics are made available The Maps and Stats WebGIS was developed to achieve objective (b), this development being undertaken primarily by officers based within the Research and Consultation Service of CBMDC, although an aspect of the capacity building activities of the BRC team has been to introduce the Maps and Stats facility as a source of relevant statistical information and to feedback users’ comments to the technical development team

In building the Maps and Stats WebGIS the intention was to provide citizens with the datasets and tools necessary to allow them to conduct their own small-area analyses and thus to develop policy and funding proposals independent of council involvement For the first time, there would be a single, comprehensive, consistent, and maintained small-area policy dataset for the district, delivered via an easily queried online system—freely available to anyone who is interested Whereas in the past, community groups would have needed to go to the council to obtain access to relevant statistics, increasingly it is envisaged that such statistics would be available directly via the Maps and Stats WebGIS The BCSP recognized that the role of the local authority, and other local agencies, as gatekeepers of local information should be lessened GIS has been criticized

as a technology that further concentrates the control of knowledge within

bure aucrac ies and exclu des relative ly disadva ntaged ci tizens (Pic kles, 1995) The BCSP explicitly aims to reve rse this trend in Bradfor d

A furth er note worthy feature of the BCSP has been the partn ership bet ween Bradfor d Council and the BRC Ghose (2001) and Ghose and Huxh old (2002) discuss pr ojects involv ing uni versity–co mmun ity partn er-ship s as a means of dem ocratizin g GIS The BCSP is bas ed upon a local gover nmen t–commu nity partn ership Rather than the coun cil devel oping

a syst em for the commu nity, the intention has been to devel op the system with the co mmun ity, with the BRC pro viding a focus fo r commun ity invo lvement Exp eriences within the project have been valuable in exp lor-ing the rela tionship s betwee n a local co uncil and its commun ities with regard to service design, provision, and use

The Map s and Stat s WebGIS (ww w.mapsa ndstat s.com) is implemen ted using AutoDesk’s MapGuide software with Microsoft’s Access as a data store Here we briefly describe some of the more innovative features deve-loped within the system

15.4.1 Boundary-Free Small-Area Estimates

A novel method for generating boundary-free small-area estimates lies at the core of the Maps and Stats system This method could have very wide application as it helps to resolve a significant problem that presently hampers small-area policy analysis

The chaotic nature of Britain’s small-area geography means that produ-cing worthwhile statistical profiles for small potential policy areas has been

a perennial bugbear for analysts Census boundaries do not coincide with postcode boundaries Health authority and police boundaries will not coin-cide with local authority boundaries and so the problems of sharing data compound Potential policy initiative areas invariably cut across data boundaries Nothing fits! And yet the increasing reliance upon evidence-based, targeted small-area antideprivation policies means that there is an increasing need for such estimates to be made

Presently there seems to be a considerable gulf between the small-area estimation methods devised by academics and those used by practitioners Small-area interpolation has attracted continuing interest from academics and a number of approaches, with varying degrees of sophistication, have been proposed (e.g., Flowerdew and Openshaw, 1987; Backen and Martin, 1989) The key point here, though, is that these research-based techniques seem to have achieved little penetration into practice As Thomasson (2000) explains, more often than not, small-area estimates within local government

are likely still to be made on the basis of simple polygon overlays or crude visual estimation

The BCSP team has devised and implemented a method of small-area estimation which is practicable within a local government context and which produces consistent, replicable results The method uses an approach that Flowerdew and Green (1991) would classify as intelligent estimation, using knowledge of the distribution of one variable, in this case residential locations, to predict others In essence, the stages of the method are:

(a) The locations of residential properties within the district were determined The starting point for this was to use the Address-Point dataset and to eliminate all nonresidential properties that were indicated within the ‘‘Organizations’’ field of the dataset This reduced dataset was then further refined by the use of high-resolution aerial photography to help identify residential areas and to exclude further nonresidential properties [See Harris and Longley (2000) and Robinson et al (2002) for further examples

on the use of remote imagery in studies of urban population distribution.] Internally generated local authority residential datasets can also be cross-matched with the Maps and Stats resi-dential properties file, so the accuracy of the file be incrementally fine-tuned

(b) The mean center of each unit postcode across the district was calculated on the basis of the identified residential properties and the number of residential properties associated with each postcode centroid recorded Standard distance calculations were used to highlight where the mean center would not be a good indicator of the location of properties within the manually reposi-tioned postcodes and centroids where necessary Sadahiro (2000) suggests that the spatial median, rather than the spatial mean, might be a more appropriate measure

(c) When any new area-based dataset is introduced into the BCSP system, a point-in-polygon operation is conducted to identify the postcode centroids which lie within each of the new dataset areas,

so constructing a postcode:data-area look-up table

(d) The dataset’s value for each area is then shared between the postcode centroids enclosed within the area, in proportion to the number of residential addresses each postcode centroid repre-sents In effect, the value associated with a data area is spread proportionately across the postcode centroids contained within the area

(e) Data are held within the BCSP system as postcode centroid esti-mates In effect, the postcode centroids become the common-pegs upon which data from disparate areal bases can be held

(f) When it is required to create an estimate for a newly defined target area, the estimated values for those postcode centroids contained within the target area can be simply aggregated All datasets are held within the system as count variables, percentages and rates only being calculated from the reaggregated target data

Figure 15.1 illustrates the process Source data are allocated to the disag-gregation points (weighted postcode centroids) and the disaggregated esti-mates are then available for reaggregation to target areas

Having experimented with this method, the BCSP is aware of some limitations in the performance of this method of estimation as it is presently implemented:

. The accuracy of the system clearly depends upon adequacy of residential address locations as a predictor of distribution of the estimated variable It is only sensible, therefore, to use the method with residentially related variables

. There are many postcodes with only a few residential properties within them In rural part of the district, i.e., that part outside the urban shape defined by the CBMDC’s Planning Department, 60% of postcodes contain less than five residential properties To avoid the instability of indicators calculated with low denominators, and also

to preserve confidentiality, the BCSP does not return results for target areas of less than 100 residential addresses This being the case, it is recognized that the estimating system is most appropriately used in built-up areas, where target areas are more likely to contain sufficient properties to rise above the 100 address thresholds

. If the source area is larger than the target area, the estimates will not

be sensitive to local conditions within the target area For this reason,

it is preferred to obtain source data on the smallest possible scale

FIGURE 15.1

Source units, disaggregation

points, and target areas.

Source unit

Disaggregation point

Target area

. The use of postc ode centr oids to represe nt the residen tial proper -ties with in a postc ode area me ans that there is pres ently some possi bility of mi sallocat ion wh en a target boun dary cuts across a uni t postcode boundary When this happ ens, all the estim ated val ue assoc iated with a postcode will be allocat ed to the target area with in wh ich the postc ode’s centr oid fall s

Further refinem ent of the method is possi ble In particu lar, the BCSP te am

is consid ering ‘‘s preading ’’ sour ce da ta directl y to individual pro perties, as this wou ld elimi nate the potenti als for error assoc iated with the use of me an center s to rep resent unit postc ode areas and thus misallocat ion prob lems where target boundari es cut postc ode a reas The initia l decisio n to use centro ids was taken in itially onl y as a pragmat ic means of red ucing pro -cessin g loads

15.4 2 User-Defi ned Target Areas

Ano ther sign ificant techni cal feature of the Map and Stat s Web GIS is that it allows users to create their own target a reas An y user can use a mo use to define , edit, and sto re any area that is of intere st to them Havi ng define d a target are a, the user then selects from a menu the da taset in whic h they are intereste d The Maps and Stat s syst em then return s a re port that provides a profile of the target area, based upon the chosen da taset Figure 15.2 sho ws the Maps and Stats interface, with a user-defined target area on screen

in this case the ‘‘Council Benefits’’ report This report provides estimates of the numbers of households claiming Housing Benefit=Council Tax Rebate

FIGURE 15.2

User-defined target area.

and Fre e Sch ool Meal s with in the user -defined target area, tog ether with the distr ict figure s fo r comparis on Note that user s can obt ain info rmation about the origins and reliabi lity of the data, and how the statistics are cal culate d,

by cl icking on links em bedded within the report page

Pro viding the publi c with the facil ity to draw their own targe t areas via the WebGIS and to retrieve smal l-area statis tics for their own define d a reas gives citizens an indepen dent ability to gene rate local- area policy proposal s Unlike man y PPGIS, the publi c are not sim ply the passive recipien ts of inform ation presen ted in a manner determined by the council Rather they have the tools to form their own views and, if they wish, to enter into a dialo g wi th the co uncil

In pr actice, the BCSP team had to balance their desire to provide open acces s to data ag ainst a proper concer n to preserve conf identi ality It might

be argu ed that, as the BCSP system return s estimate s rath er than actual observ ed da ta val ues, the issue of conf identi ality does not arise The view was take n however , that mo re stringen t restra ints shou ld be put in place The syst em has been desig ned so that it will not return profiles for user -defin ed areas that contain less than 100 res idential addresse s

15.4 3 Inform ation Dissem ination Rather Than Analysi s

As desc ribed ab ove, the majo r outputs from the syst em at present are stati stical tabu lations profilin g user-de fined areas The system does contain some preprepared thematic maps of selected variables, and because of the work done to identify residential locations for the estimation routine, these can be presented in dasymetric, rather than conventional choropleth for-mats Figure 15.4, for example , shows the Maps and Stats site interf ace with

FIGURE 15.3

A target area specific report.