New Ways of Mapping Social Inclusion in Dublin City

A Joint Initiative of Dublin City Partnership/Dublin City Development Board

Rob Kitchin and Justin Gleeson

Oak Room, Mansion House, 14th May 2009

Presentation Outline

• Project background

• Objectives of the project

• The Evidence Base – The Present Situation

• Improving the Evidence Base – The Pilot Project

– 4 Case studies

• Project Recommendations and Next Steps

Project Background

• In March 2008, PLANET – The Partnerships Network, in conjunction with the SIM group of Dublin City Development Board and Dublin City Council engaged NIRSA to:– Explore new ways to spatially analyse and map social inclusion data in

Dublin City

• The rationale for the project was that providing partnerships and others concerned with social inclusion with a greater level of detail and access to information at a local community level would:– Enable better area-based policy formulation based upon a fuller

understanding of the nature of local communities and their epecific needs

• Having a detailed and robust evidence base as a background to policy development will ensure maximum benefit is drawn from public finances

• One way to ensure this is the case is to map relevant data at as finer resolutions as possible to expose the patterns that characterise an area and its population.

• With a specific focus on Dublin-based partnership areas a pilot project was undertaken in two study locations– Ballyfermot/Chapelizod Partnership

– Northside Partnership

Objectives of the project

1. Identify data relating to social inclusion held by various data and service agencies, including those that do not traditionally release data or do so at course spatial scales.

2. Work with Dublin City Development Board to persuade agencies to release data at finer scales than usual and where necessary to geo-code data for analysis and mapping in the pilot locations.

3. Structure that data into a coherent database for analysis and mapping, and to work with Dublin City Development Board to explore the possibilities of adding any geo-coded data that might possess (e.g. service locations).

4. Examine various established deprivation indexes (from Ireland and abroad), explore new possible indexes, and to test their usefulness and validity in a Dublin context.

5. Map, where possible, data and indexes into existing administrative boundaries – Dublin City, Partnership, ED and EA.

6. Map various data into new boundaries, specifically the new Small Areas created by National Centre for Geocomputation (NCG) for Ordnance Survey Ireland (OSi).

7. Explore ways of accessing and presenting the data for non expert users, including the development of online mapping tools.

The project was undertaken over a one year period from March 2008 to April 2009 with regular update meetings with the steering committee:

Ballyfermot/Chapelizod Partnership, Northside Partnership, Dublin City Development Board and Dublin City Council Social Inclusion Unit.

The Evidence Base – The Present Situation

• Traditionally the evidence base has consisted of the census data delivered to partnerships at the Electoral Division (ED) level– This has helped inform policy making and the delivery of

programmes– It can be significantly improved in two main respects

• First, since 2002 the CSO have released statistical data at a more detailed spatial scale – the Enumeration Area (EA)– 5 county borough areas (Dublin, Cork, Limerick, Galway and

Waterford)• Secondly, there is potentially significantly more data that can

be used as part of the evidence base • Allying census data with health, welfare, employment and

service provision data will provide a much richer understanding of a population within a locale– Collected at a great frequency– Can be used to monitor changes over time – monthly, quarterly or yearlt basis

Electoral Divisions (EDs)

• 3,441 legally defined EDs• Census data not available for all EDs. 2006 census output for a national coverage of 3,409 EDs• Advantage: stability of boundaries and longetivity of use

• Highly variable in spatial size and in population size• Rural EDs can be < 100• Urban can be > 20,000 (Dublin average = 3,570)• Banchardstown-Blakestown has a population of 32,288• The larger the population within an ED the more that variation amongst the population is masked through the effects of aggregation

The Statistical Geography of Ireland

Enumerator Areas (EAs)

Enumerator Areas (EAs)

• EA areas constitute the are/workload assigned to each enumerator• geographical extent is restricted to approx 330 households

• much finer spatial resolution• less masking of local variation

• EA data is avilable for the 2002 and 2006 census (5 county borough areas)• No ability to undertake a time series analysis as the spatial units are not stabe and change over time• EAs are only released for county boroughs and suburbs, not all partnerships in Dublin have a complete coverage

• Blanchardstown, CPLN, Dodder Valley and Southside do not have full EA coverage

The Statistical Geography of Ireland

Small Areas (SAs)

Small Areas (SAs)

• The creation of the new Small Areas has been undertaken by the NCG at NUI Maynooth on behalf of OSi• Each SA should contain a minimum of 65 households• SAs should nest into townland, ED and county boundaries

• There will be approx 17,000 SAs covering the whole country when the boundaries are released for use• CSO will be releasing the 2011 Census at the new SA level, it is also hoped that the 2006 Census will be back-fitted to these new spatial units

Geography Planet Partnerships Ballyfermot Northside

ED 292 8 40

EA 1107 26 136

SA (Beta) 3636 81 425

Availability of Geo-referenced data

• The census Small Area Population Statistics (SAPS)– ED level data provided to partnerships through the administering body of

Local Development and Social Inclusion Programme (LDSIP)

– 9 sub-themes with tables detailing the relative strengths and weaknesses of variables across the partnership EDs

– Raw figures and percentage format

– Context with National, Regional and County figures

– Majority of the data is available for 2006, a number are also provided in a time-series manner (1996-2006)

– Static maps at partnership and Dublin City scale

– Partnership maps created using the national data distribution range• Enables a useful comparison to the national profile• Can also mask local internal partnership variation• It is useful to map each partnership on an individual basis, even more important

when using finer spatial scale (increased variation)

Availability of Geo-referenced data

• The New Measure of Deprivation– Partnerships have access to the Haase&Pratschke Index of Deprivation

through the LDSIP administrators

– Provides an analysis of change in deprivation in Ireland from 1991 to 2006

– Using ED data the index is based on 3 dimensions• Demographic profile

• Social Class composition

• Labour market situation

– This has also been extended to EA level using a combined ED and EA dataset (Dublin Inner City Partnership)

– Small Area Health Research Unit (SAHRU) at Trinity also produce a national deprivation index at ED level

– Intended to reflect material (not social) deprivation

– Unemployed, low social class, car availability, local authority housing

Availability of Geo-referenced data• Place of Work Census of Anonymised Records (PoWCAR)

– origin and destination of workforce

– An analysis of this dataset can provide partnerships with a detailed profile of the work interactions of the community

• What are the main employment locations for a partnerships residents?

• How accessible are employment locations to the local community?

• Are certain areas of the partnership more dependent on a specific type of employment?

Availability of Geo-referenced data

• Public Facility Data– Local Authorities in Ireland have a geocoded inventory of public facility

data

– Such data can be used to calculate the level of access for populations within partnership areas to key services

• GPs• Primary schools• Secondary schools• Post offices• Accident and emergency units• Training and empoyment agencies etc

– Access to services is often used in indices of deprivation but at present such data is not widely used to access social inclusion within Ireland.

Availability of Geo-referenced data

• Live Register Data– Extracts for the Live Register at the individual office level are available to

those interested in monitoring social inclusion on a weekly and monthly basis

– Provides an indication of the welfare trends in an area but there is no specific catchment assigned to the individual social welfare offices

• For instance, extracts from the Ballyfermot office can also include persons residing in other partnerships such as CPLN

– As a result, understanding the spatial distribution of such data is difficultBallyfermot

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

Jan-

05

Apr-05

Jul-0

5

Oct-0

5

Jan-

06

Apr-06

Jul-0

6

Oct-0

6

Jan-

07

Apr-07

Jul-0

7

Oct-0

7

Jan-

08

Apr-08

Jul-0

8

Oct-0

8

Month-Year

% C

han

ge

fro

m A

nn

ual

Ave

rag

e

?

Improving the Evidence Base – The Pilot Project

• The overall objective of the pilot project was to explore ways to improve the evidence base of the area-based partnerships in six main respects– To make available data and analysis little used with respect

to deprivation

– To generate and analyse new kinds of spatial data

– To increase the spatial resolution of analysis

– To examine ways to make the data and outputs easily available to non-expert users

– To illustrate the utility of an improved evidence base

• To that end the project was divided into four phases, each examining one or more of these objectives;

– Creating a census atlas of Dublin at ED and EA scales

– Using POWCAR data to examine labour markets

– Examining deprivation indices and access to services

– Utilising the Small Areas to examine geo-coded welfare data

• Each of these phases involved fundamental research into the nature of the data and issues at hand, followed by a case study applying and testing the knowledge gained with respect to the pilot areas of Ballyfermot/Chapelizod and Northside partnerships

1. Creating a census atlas of Dublin at ED and EA scales• Produced a series of maps for Dublin city and for

individual partnerships at both ED and EA level• ED data available for 1991,1996, 2002 and 2006

– 80 census variables have been mapped and made available to both Ballyfermot/Chapelizod and Northside partnerships

– Instant Atlas interactive module

• EA scale provides a much more detailed spatial resolution of the trends at a local community level– Ballyfermot/Chapelizod = 8 EDs– Ballyfermot/Chapelizod = 26 EAs

• 80 EA variables available (2006 only)

Theme # of Variables

ED Year EA Year

Population (raw) 21 1991,1996,2002,2006

2006

Population (percentages) 21 1991,1996,2002,2006

2006

Economic Status (raw) 9 1991,1996,2002,2006

2006

Economic Status (percentages) 9 1991,1996,2002,2006

2006

Education (raw) 10 1991,1996,2002,2006

2006

Education (percentages) 10 1991,1996,2002,2006

2006

Social Class (raw) 8 1991,1996,2002,2006

2006

Social Class (percentages) 8 1991,1996,2002,2006

2006

Socio-Economic Group (raw) 24 1991,1996,2002,2006

2006

Socio-Economic Group (percentages) 24 1991,1996,2002,2006

2006

Housing Tenure (raw) 4 1991,1996,2002,2006

2006

Housing Tenure (percentages) 4 1991,1996,2002,2006

2006

Car Ownership (raw) 5 1991,1996,2002,2006

2006

Car Ownership (percentages) 5 1991,1996,2002,2006

2006

• EA data is available in Instant Atlas

• Created city level maps for 12 variables with particular resonance to deprivation– Population Density

– % Population 0-14 Age Band

– % Population 65+ Age Band

– Age Dependency Ratio

– Unemployment Rate (Total, Male and Female)

– Highest level of Education (No Formal/Primary)

– Car Ownership levels

– % Socio-Economic Group: skilled and unskilled

– Housing Tenure: Buying or Renting from LA

– Social Class: skilled and unskilled

• Whilst static census maps are useful to partnerships one of the prime objectives of the pilot project was to explore ways of making data more accessible and interactive

• Evaluated a number of online mapping packages• Two interactive censuses were produced, one for

Ballyfermot/Chapelizod and one or Northside– ED for Northside– EA for Northside

www.planet.ie /mappingmodule/northside

CSO Data / OSi boundary (ED/EA/SA) / Instant Atlas software

.html file for each Partnership1 licence1 boundary file

Case Study 1: Enumerator Area Analysis in Northside Partnership

• Designed to highlight the utility of conducting analysis at the EA scale, as opposed to ED scale

• Preparing for Life Programme– Early intervention plan that

provides support for parents

– Children become better prepared for school by age 5

• Originally rolled out for Priorswood B and C

• Proposed extension to D and E– Demographic analysis

– Unemployment rates, lone parents, Education levels

• Expert panel were satisfied to include Priorswood D but not Priorswood E based on criteria

• Counter to in-depth street level knowledge of partnership– Glin Area

• Undertake analysis at EA level

• Glin area is more disadvantaged than other areas of Priorswood E– Included in PLF

0.00

10.00

20.00

30.00

40.00

50.00

60.00

081 PriorswoodB

082 PriorswoodC

083 PriorswoodD

084 PriorswoodE

ED

%

Unemployment Rate

Lone Parents

Low Education

0.00

10.00

20.00

30.00

40.00

50.00

60.00

081Priorswood

B

082Priorswood

C

083Priorswood

D

084Priorswood

E

02/429

ED/EA

%

Unemployment Rate

Lone Parents

Low Education

2. Using the POWCAR data to examine labour markets

• A key aspect of deprivation is access to waged employment

– little spatial analysis of in Ireland of access to work at highly localised scales

• POWCAR raw dataset = 1.8m records

– Simplified to create ED to ED/EA interactions

• Socio-economic group, industrial group, means of travel, levels of education etc

• Two excel databases focussing on travel to work interactions within and out of each partnership.

• Series of ED maps detailing the commuting patterns for each partnership– SEG, Mode of Transport etc

Case Study 2: POWCAR analysis in Ballyfermot – Public Transport Access to Blanchardstown and

Mulhuddart Job Pools

• Partnership access to large labour pools in– Liffey Valley– Blanchardstown/Mulhuddart

• Both outside the partnership area but are regarded as having significant employment potential

• Liffey Valley was well served with public transport• Partnership felt that the lack of adequate direct public transport to

Blanchardstown was acting as a barrier to employment opportunities• No evidence to support this!

47% of partnership workforce employed in Liffey Valley use Public Transport

11% of partnership workforce employed in Blanchardstown/Mulhuddart use Public Transport

3. Examining deprivation indices and access to services

• Partnerships receive the Haase&Pratschke Index at ED level– Partially updated using ED and EA for Dublin City in 2006

• Extended the SAHRU Index to the EA scale for Dublin alone– Negates some of the effects of averaging at the ED scale

• Examined deprivation indices used by other countries

– N Ireland, England, Scotland etc

• Many similarities with Irish indices but also some key differences - use of non census data– Incidence of crime

– Employment/unemployment and benefit data

– Access to services (GPs, Schools, Supermarkets)

• Lack of geo-coded data in Ireland, extremely varied and sub-standard levels of address collection within government agencies

• Presently difficult to include any non-census variables in any index of deprivation

• Even use them individually as a measures of social inclusion

Rep of Ireland Northern Ireland Scotland Wales England

Hasse&Pratsche SAHRU

Demographic Decline (4)

Unemployment (1)

Income (4) Income (5) Income (5) Income (2)

Social Class Disadvantage (3)

Low Social Class (1)

Employment (3) Employment (4) Employment (4) Employment (4)

Labour Market Deprivation (3)

No Car (1) Health (5) Health (7) Health (4) Health (4)

Rented Accommodation (1)

Education, Skills and Training (6)

Education (5) Education, Skills and Training (4)

Education (4)

Proximity to Services (8)

Access to Services (8)

Access to Services (4)

Access to Services (4)

Housing (3) Housing (2) Housing (3) Living Environment (4)

Social Environment (8) Crime (5) Crime (4)

Case Study 3 – Access to Public Services

• Key data– Residence/population (ED/EA/SA or Residential point)– Public facility (GPs, Schools, A&E, Shopping facilities etc)

• Data available from Dublin City Council– Public facility database

• Wide variety of different techniques can be used to measure basic accessibility to key services such as:– Road distance (geographical unit centroid) to nearest facility– Road distance (every residential address point) to nearest facility– Euclidean (crow flight) distance to nearest facility– Public Transport distance/time to nearest facility

• Time of departure• Train/bus timetables

Distance (crow flight) from every residential address point to nearest

set of facility (GP, Primary and Secondary Schools)

Drive-time distance from every residential address to the

nearest facility (GPs)

4. Utilising the Small Areas to examine geo-coded welfare data

• Government departments routinely collect data about their work and the constituitents they serve

• Much of these data are potentially of use in the monitoring of social inclusion

– Welfare, health, education and housing data

• Poorly geo-coded and typically only made avilable at national, regional or county levels

• Primary objective of the pilot project

– Encourage government departments to make such datasets available for analysis at EA and Small Area level

– Highlights data’s utility with regard to their own work and policy formulation

• Essential work if Ireland is to become compliant with the EU INSPIRE directive (conform by 2014)

– Improve data collection, address format protocols, access and use of data

Case Study 4 – Dependence on the Social Welfare System in Ballyfermot

• Department for Social and Family Affairs– Short-term Live Register data

– Dublin 10 & 20

• Within each extract a number of key variables were available– Claim code (JB, JA, OPF, BtW etc)

– Claim Start date

– Prev Occupation

– Child dependents #

– Personal rate

– Child dep rate

– Total payment

– Gender

– Date of Birth

– Marital Status

– Country Code

• Address Information– House Number

– Street Name

– Local Area

– Postcode

• Address information swapped for Geographical reference– ED Code, EA Code, Small Area Code

ID_2642 # Ballyfermot Road, Ballyfermot, Dublin 10

ID_2642 Decies, 02072, 268057008, Ballyfermot, Dublin 10

Geo-coding without House #

Geo-coding with House #

• Address matching being undertaken after data has been collected by department

• To geo-code accurately and efficiently the address must match GeoDirectory (national address database)– Cleggan Road : Bothar an Chloiginn

• Cleggan Road• Chloiginn Road• Cloiginn Road• Cloigann Road• Cloighann Road• Cloighinn Road

• Need to assign applicants address information to listing from national address database– To be done at data entry stage– Address look up– Easy match to ED,EA,SA– Data could be extratced to match any boundary

• Partnership, Garda, Health etc

Ballyfermot

-10.00%

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

Month-Year

% C

han

ge

fro

m A

nn

ual

Ave

rag

e

Geo-coded and mapped to ED

Geo-coded and mapped to EA

Geo-coded and mapped to SA

Increase in numbers on the Live Register from w/e 08/08/08 to w/e 20/02/09

Increase in numbers on the Live Register (JB) from w/e 08/08/08 to 20/02/09

Recommendations1. Conducting a feasibility conducting a feasibility and cost benefit

analysis of providing the data to all partnerships and other parties using a user-friendly, interactive, web-based system such as Instant Atlas

2. Working with government departments to examine how their data can be geocoded and mapped at the Small Area scale.

3. Extending the analysis to rural areas to ensure that the methods developed can be rolled out systematically for the entire state;

4. Extracting as much value as possible from existing datasets that are at present often overlooked;

5. Examining the technical and ethical issues that may arise in mapping the 2006/2011 census data at the Small Area scale

6. Conducting more research into the development of a deprivation index that extends beyond census data to include measures of access to services and other data relating to health, housing, crime, education and so on;

7. Exploring whether it would be possible to undertake the analysis on an all-island basis

Next Steps• Development of a strategy to ensure that these new

forms of mapping are widely available to those who need it.

• A set of Pilot Projects examining– the geo-coding of data from collection to output within a

government department, including a cost benefit analysis

– the extension of the techniques developed to rural areas

– The development of a deprivation index including access to services and non-census data

– Cost-benefit analysis of development of a web-based system for delivery of geo-coded data to the public sector

– the creation of a national geo-coded public facility database