Data Real Estate

reviewed paper

Where are the poor? A disaggregation approach of mapping urban poverty

Germene STEWART, Monika KUFFER

Masters Urban Planning & Management Germene Stewart, Central Housing and Planning Authority, Planning and Settlement

Development Department, Lot 237 Camp Street, Cummingsburg, Georgetown, **************@*****.***

Dipl.-Geogr. Monika Kuffer, International Institute for Geo-Information Science and Earth Observation, ITC, Department of Urban

and Regional Planning and Geo-information Management; PO Box 6, 7500 AA Enschede, The Netherlands; ******@***.**

ABSTRACT

In this paper, we highlight an approach for mapping indicators of poverty. The main drive for developing

such an approach is to address the neglected aspect of the spatial dimension of poverty since in most cases

the indicators chosen for poverty rely on household data. Such data if spatially analyzed are usually

aggregated to administrative units such as neighbourhoods or wards. Therefore the spatial heterogeneity

within the administrative unit is hidden. For target interventions, detailed spatial information is needed on

poverty as a multidimensional phenomenon at a disaggregated level.

The study shows an approach to address the neglected aspect of the spatial dimension of poverty, based on a

case study of the city of Cebu in the Philippines. It is integrating locally available datasets on poverty

relevant issues from several municipal departments with information extracted from a recent very-high

resolution satellite image. The indicators identified and mapped are dwelling size, building density, lack of

proper road network, poor structural quality of buildings and access to water and sanitation. Analysis is done

at both disaggregated and aggregated levels (using data from the census) to demonstrate how data

aggregation can hide spatial variation of poverty but also to examine the robustness of the selected spatial

indicators.

The results at the disaggregated level clearly locate poverty areas and show that the nature of poverty varies

among those areas. For some areas, the main issue is e.g. structural quality of buildings, while for others it is

either access to proper sanitation and/or access to water supply. This approach has potential by using locally

available datasets in combination with satellite image data to provide spatially detailed information usable

for both planning and monitoring poverty intervention.

1 INTRODUCTION

To eradicate extreme poverty is one of the Millennium Development Goals (MDG s), the target is to halve

the proportion of people whose income is less than one dollar a day, by 2015 (UN/DESA, 2006). Presently,

more than a billion people in the developing world live on less than one dollar a day, with an increasing

number and proportion of poor people living in urban areas. Even though efforts are made to localise urban

poverty in many developing cities, the scarcity of relevant data, coupled with lack of both human and

financial resource for data collection and analysis is a significant constraint to target interventions.

Traditionally approaches of measuring poverty have been focusing on income and consumption (monetary

dimension of poverty), while recently poverty is understood as a multidimensional phenomenon focusing on

multiple sources of deprivation in poverty areas (Martinez-Martin, 2005). Those areas are characterized by

overcrowding, insufficient water supply, sanitation and infrastructure, problems of health and nutrition,

limited access to education, as well as insecurity, exposure to hazard, deficient social relations, etc (Turkstra

and Raithelhuber, 2004). As a multidimensional phenomenon urban poverty is also spatially heterogeneous

since poor people tend to be clustered in specific places. Therefore it is important to capture the spatial

heterogeneity and the substantial variation of poverty areas to know better where the poor are and how they

are distributed throughout the city. Targeting of the poor depends significantly on how poverty is

conceptualized, measured and analyzed within a specific local context. To be able to locate the urban poor

and adequately profile them for target intervention requires defining, identifying and understanding areas of

poverty not only on their commonalities but also on their diverse characteristics in terms of social, economic

and physical conditions in a local context.

In principle, there are three main approaches of analysing poverty depending on the availability of data

(Zeller et al., 2006), namely the construction of a poverty line, rapid appraisal and participatory appraisal

methods or construction of a weighted poverty index (combining qualitative and quantitative indicators).

REAL CORP 007 Proceedings / Tagungsband ISBN: 978-39502139-2-8 (CD-ROM); ISBN: 978-39502139-3-5 (Print)

847

Vienna, May 20-23 2007 www.corp.at Editors: Manfred SCHRENK, Vasily V. POPOVICH, Josef BENEDIKT

Where are the poor? A disaggregation approach of mapping urban poverty

In most countries if spatial analysis of poverty is done it is at national, city or district level. This aggregation

of data hides the stark contrast of income and living conditions between better-off urban citizens and the

urban poor and overlooks pockets of poverty areas within the city (Hall et al., 2001). In recent years, poverty

mapping, which is a powerful new information analysis and communication tool, has come into prominence

and is bridging this gap in provision of the required spatial information. A variety of methods to spatially

locate poverty has been developed e.g. small-area estimation (combining census and survey data for

disaggregated geographic units such as municipalities), multivariate weighted basic-need index

(disaggregating poverty using statistical techniques such as principal components), combination of

qualitative information and secondary data (e.g. mapping poverty based on livelihood strategies),

extrapolation of participatory approaches (incorporating local perception of poverty), direct measurement

using household-survey data or census data (Davis, 2003).

One recent study presented by Baud et al. (2006), shows the relevance of a mapping urban poverty on sub-

city level in Delhi (India), comparing the different wards by conceptualizing poverty by an index of multiple

deprivation. The potential of recent very-high resolution satellite images (providing spatial resolutions of less

than one meter) to disaggregate poverty further than to administrative levels of neighbourhood or wards has

been explored by Lemma et al. (2006). Very-high-resolution satellite images supported by local knowledge,

field observations and available local data can provide information on the location of poverty areas. The

integration of these datasets in a GIS can generate a wealth of information on urban poverty. The advantage

of such an approach is that it is locally generated, and thus easier for institutional embedding. Further, the

use of very-high resolution images as backdrop images for putting the poor on the map, is an important

aspect for gaining political support and communicating results to non-technical audiences at the local level

(Turkstra and Raithelhuber, 2004).

2 BACKGROUND

Cebu City is located on the central eastern part of Cebu Province, an island at the centre of the Visayas in

Southern Philippines and is the capital of the province as well as of the Central Visayas region. Based on the

last population census conducted in 2000, the city had a total population of approximately 700,000. As

Figure 1 shows, Cebu City is subdivided into 80 barangays (smallest administrative unit, similar to wards),

49 urban barangays where the majority of the population lives and 31 rural barangays (the later study focuses

on 12 centrally located urban barangays).

High development pressure as a consequence of economic growth during the last decade has dislocated the

urban poor into overcrowded and hazardous areas. It is pertinent to note that the concept of poverty in Cebu

is viewed differently by the urban poor and the policy and decision makers. According to the study

conducted by Etemadi (2001), the urban poor view poverty as deficiency in income, living conditions, access

to basic services and resources, and working conditions. While policy and decision makers view the urban

poor as the homeless and landless even if they have earnings. As a consequence the main thrust of the

policy and decision makers, is providing services and assistance whenever there is an incidence of

demolition (Etemadi, 2001).

REAL CORP 007: To Plan Is Not Enough: Strategies, Plans, Concepts, Projects

848 and their successful implementation in Urban, Regional and Real Estate Development

Germene STEWART, Monika KUFFER

Figure 1: Population Density in Cebu

The Division for the Welfare for the Urban Poor (DWUP) has the mandate to provide adequate housing

facilities, such as the socialized housing facility to the urban poor and ultimately relocate them from hazard

zones of the cities. The poor however are required to organize themselves into cooperatives associations

known as the Homeowners Associations (HOA). There are about 600 of these Homeowners Associations in

Cebu City. The DWUP usually stimulates the formation of these associations in order to provide them with

assistance in the form of:

Assisting the HOA in case of eviction actions taken by landowners.

Buying land for the HOA, for onsite settlement or land resettlement (mostly into fringe areas). In

these cases the households are required to pay for the land in installments.

Playing a leading role in providing basic services to the sites acquired.

Most of the poverty areas in Cebu are located on low-lying pieces of land squeezed between subdivisions

and factories in highly polluted industrial areas. In some cases the areas are so crowded that some dwellings

have been extended unto the water surface along the coastline. In most of the areas basic services like

potable water supply, sanitary facilities and proper road network are almost non existent. The structural

quality of most dwelling units is poor and can be deemed shanties.

In general there is tremendous lack of public

spaces. Poverty areas have very high population and building densities. The density is usually greater than 60

households per hectare. Plot and dwelling size is usually below the minimum requirement of 36 and 25

square meters respectively. Detached buildings do not comply with the legal set back requirements of 1.5m

set back side distance and 3m set back front distance. However, recognizing that urban poverty is a major

problem facing the city, the Cebu city government has become one of the most active local government units

in the Philippines dealing with this phenomenon.

Characteristics of poverty areas (pockets of poverty) in Cebu:

Located on low-lying land (unto water surface) and near industrial areas.

Overcrowded areas with high densities with lack of public space.

Plot and dwelling size below the minimum requirement.

Insufficient basic services - potable water supply, sanitary facilities, roads.

Shanties makeshift structures ( barong-barong ) with non compliance to building regulations.

CEIT ALANOVA

849

Central European Institute of Technology,

Dept. for Urbanism, Transport,

Environment & Information Society

Where are the poor? A disaggregation approach of mapping urban poverty

Figure 2: Characteristics of poverty areas in Cebu

3 METHODOLOGY

Acquiring comprehensive information on the spatial location of pockets of poverty, reflecting their

heterogeneity requires the integration of multiple data sources. We employed group discussions with local

experts and stakeholders to develop a list of indicators of locating pockets of poverty, as well as obtaining a

good overview of locally available data to support their implementation. The group discussions were held

with the Chief Planning Officer and other senior personnel of the Planning Department, the coordinator of

Division of Welfare for the Urban Poor, a senior official of the National Statistical Office and the community

leader of the Basak San Nicholas Sitio Pandayen Homeowner Association. Based on the group discussions

the following list of indicators for identifying pockets of poverty has been developed (see table 1).

REAL CORP 007: To Plan Is Not Enough: Strategies, Plans, Concepts, Projects

850 and their successful implementation in Urban, Regional and Real Estate Development

Germene STEWART, Monika KUFFER

Table 1: Disaggregated indicators of poverty pockets

The first group of indicators is the basis for identifying pockets of poverty and the second group (living

conditions) is reflecting the diverse characteristics poverty. Optimally more indicators could have been

developed for reflecting the diversity of poverty (using e.g. social dimension) but the employed indicators

are understood as examples to combine spatial disaggregation with an analysis of poverty as a

multidimensional phenomenon. For identifying the visible-morphological characteristics the main data

sources was a very-high resolution satellite image (QuickBird 2005). However, since not all the

characteristics of these areas are visible from space, data have been collected from other sources, namely

flood records, location of HOA and dry hydrants. Field observation was also an important source since it

could validate the data collected from the other sources. Data used for analysing differences in living

conditions of the identified pockets of poverty was a dataset containing information regarding water

connection and the availability of toilet facility of households (Watsan4).

The indicators were compared and analyzed with two indicators (namely dwelling size and structural quality

of dwellings) derived from the census database (2000) on aggregated data at barangay level to determine

whether there is any correlation. The statistical data were used to determine whether the barangays with a

larger coverage of poverty pockets (at disaggregated level) have also a higher number of small buildings

(below the minimum requirement) or higher number poor structural quality of buildings.

CEIT ALANOVA

851

Central European Institute of Technology,

Dept. for Urbanism, Transport,

Environment & Information Society

Where are the poor? A disaggregation approach of mapping urban poverty

3.1 Identifying pockets of poverty

Pockets of poverty can be located on a QuickBird image by describing typical characteristics based on

colour, texture, size and pattern (Lemma et al. 2006). Based on these interpretation elements the

characteristics of the poverty pockets visible on the image, are:

Irregular layout pattern inability to observe formal road network within settlements and non

visibility of access roads or footpaths,

Extreme clustering and continuous roof coverage of buildings (texture), greater than 90% roof

coverage,

Small size of buildings of less than

the stipulated required 25 m,

Tabl e 2: Image interpretation elements

Bluish-grey appearance of roof materials of the clustered buildings (colour).

These visible characteristics were used as criteria in identifying the poverty pockets. They are easily

observable on a satellite image depending on the display scale. At different scales, the indicators have their

strengths and weaknesses. For instance, the characteristics of colour and texture the bluish-grey appearance

of roof materials and the extreme clustering of buildings are good indicators at a scale of 1:8,000. On the

other hand, the characteristic of pattern irregular layout and non visibility of access roads or footpaths are

better indicators at a large scale such as 1:1,200.

REAL CORP 007: To Plan Is Not Enough: Strategies, Plans, Concepts, Projects

852 and their successful implementation in Urban, Regional and Real Estate Development

Germene STEWART, Monika KUFFER

Figure 3: Comparison of indicators at different scales for a poverty pockets in Cebu (QuickBird 2005)

The small size of buildings tends to be a fuzzy indicator at both scales. At the smaller scale small buildings

are not clearly visible on the satellite image. At the larger scale it is somewhat difficult to clearly distinguish

pockets of poverty pockets based on small size of buildings since no areas can be classified as totally

homogeneous. Any particular block or area within the city contains buildings with different sizes.

Considering these strengths and weaknesses, visual interpretation was therefore be done at two levels of

visibility, one at a scale of 1:8,000 and then at a larger scale of 1:1,200.

The delineation of the poverty pockets was done by on-screen digitizing. To facilitate digitizing, specific

interpretation guidelines were employed. Firstly, the minimum size of a poverty pockets had to be

established. In exploring the Homeowner Association (HOA) records, it was found that they have none less

than 7 structures for one specific location. Using this as reference in combination with the average dwelling

size of 14 m, a minimum of approximately 100 m was established as minimum mapping area. Secondly, in

order to assist with the interpretation, available data layers of the major roads, rivers and barangay boundary

were overlaid on the satellite image.

Not all spatial indicators are visible on the image, thus further supporting indicators have been used on e.g.

the location of hazards zones. For the indicators not visible from the satellite image, data on these indicators

is extracted from the following datasets:

Flood prone areas: Point locations of previous flooding events, such areas are mostly occupied by

the poor, since they tend to occupy land that is prone to environmental hazards such as floods.

Dry hydrants: all records of outlet points. These outlets are located in very high density areas

inaccessible to emergency vehicles such as the fire engines. Such areas are mostly areas with poor

living conditions.

Location of HOA: The HOA capture the poor that organised them in Homeowner Association.

3.2 Analysing diversity of poverty pockets based on living conditions

In order to analyse differences between the identified pockets of poverty two measures that indicate

differences in living conditions were selected, namely access to water and sanitation. The data were retrieved

from the Watsan dataset. The records enable to identify households with no toilet facility and records with

water source far from house, e.g. open wells whereby households must get water by container, this type of

water source has also possible contamination. They are classified as households without piped water.

Notwithstanding the possible geometric inaccuracy5 of the point data selected, the clustering of the points at

CEIT ALANOVA

853

Central European Institute of Technology,

Dept. for Urbanism, Transport,

Environment & Information Society

Where are the poor? A disaggregation approach of mapping urban poverty

previously defined pocket of poverty (on the QuickBird image) give evidence of two aspect of the diversity

of living conditions between the pocket of poverty.

3.3 Census data to indicate poverty levels on aggregated units

In order to analyse the robustness of the selected indicators for identifying pockets of poverty, there was

need to compare the results with available data from the census that provide information on poverty. Since

dwelling size and structural quality of buildings were selected previously as indicators, data on these

indicators were accordingly extracted from the census. The census data is aggregated at the level of the

barangays. The data on dwelling size was available in interval range and not by absolute numbers. Therefore

all records of dwelling size were selected for the barangays under study within the interval ranges of 0-9 m

and 10-19 m . Records of dwelling size within the interval range of 20-29 m were not selected since these

will include dwellings above the minimum requirement of 25 m, which is not defined as a characteristic of

poverty pockets.

When overlaying the data on the QuickBird image the points did not accurately overlap with the buildings

visible on the image.

4 DISCUSSION OF RESULTS

For implementing the set of indicators, 12 centrally located urban barangays were selected (see figure 1).

Based on the interpretation guidelines suspected areas of poverty were delineated. But, visual image

interpretation could not determine ultimately whether these areas are occupied by poor households. Further

evidence was needed to verify this classification, using locally available data sources. The location points of

HOA, dry hydrants, flood prone areas greatly facilitated data extraction of the pocket of poverty. Figure 4

illustrates the clustering of the locations of the Homeowners Associations, the dry hydrants and the flood

prone areas and locations identified from the image interpretation using the morphological indicators.

Figure 4: Pockets of poverty combining results of the visual interpretation with locally available data sources

REAL CORP 007: To Plan Is Not Enough: Strategies, Plans, Concepts, Projects

854 and their successful implementation in Urban, Regional and Real Estate Development

Germene STEWART, Monika KUFFER

Figure 5: Finally extracted pockets of poverty

The selected data (figure 4) represent characteristics of the poverty pockets in terms of their location, density

and vulnerability to environmental hazard. The data was then filtered to select only those areas equal or

greater than 100m . As a consequence some areas on the map (see figure 5) where there are no poverty

pockets, but there are some clustering of data, are those polygons of less than 100m . The total number of

poverty pockets is 84 and accounted for 52.5 ha (12%) of the total land area of the barangays under focus.

The smallest area is 0.01 ha while the largest is 3.3 ha.

For further analysis at a disaggregated level, the data were integrated with the Watsan data since this data is

at the level of households and is able to provide both spatial and aspatial information on households

characteristics. The major limitation of using the Watsan data was the time difference and incompleteness of

the dataset (two barangays were not covered by the dataset). In analysing the issue of water and sanitation

the Watsan data was extracted according to households:

Households (HH) without piped water.

Households (HH) without toilet.

This distinction had to be made to give a better overview of the living conditions in the identified poverty

pockets, using the important aspect of water and sanitation. Analysing the first category, the percentage of

households without piped water located within the poverty pockets is 16%. The poverty pockets with the

most households without piped water were found in the barangays of Basak San Nicholas and Mambaling

with 88% and 38% respectively (see table 2).

Table 3: Poverty pockets with most problematic living standard

CEIT ALANOVA

855

Central European Institute of Technology,

Dept. for Urbanism, Transport,

Environment & Information Society

Where are the poor? A disaggregation approach of mapping urban poverty

Figure 6: a) Households without piped water b) Households without toilets in poverty pockets

In analysing the second category which is based on proper sanitation, the percentage of households without

toilet located within the poverty pockets is 25%. Comparing with the results of the first category the problem

of sanitation is more widespread. The poverty pockets in Mambaling and Pasil have the highest percentages

with 98%, 74% and 66% respectively (see table 2).

At an aggregated level of the barangay comparison with the census data for some indicators used at

disaggregated level is possible. The poverty polygons were summarized according to barangays and the %

coverage of poverty pockets was then calculated for each barangay to determine which barangay has the

highest coverage of poverty pockets.

Table 4: Comparing aggregated with disaggregated data

The comparison of poverty levels in table 3 based on the census data and the located pockets of poverty

show a moderate correlation of 0.5 in both cases ( % of small buildings and % of makeshift materials

compared with the % of poverty pockets ). The moderate correlation of the data can be explained by several

reasons, from the census data only dwelling of less than 20m were derived while using for the visual

interpretation 25 m . Another reason can be attributed to the time difference of 5 years (between the census

and the satellite image). For the barangays of San Nicholas Proper and Pahina San Nicholas no poverty

pockets have been delineated, while the indicators derived from the census point to poverty which is not

spatially clustered. Poverty is most pronounced in Ermita, having 27% of the area delineated as pockets of

poverty as well 53% of substandard housing size and 15% substandard housing quality.

REAL CORP 007: To Plan Is Not Enough: Strategies, Plans, Concepts, Projects

856 and their successful implementation in Urban, Regional and Real Estate Development

Germene STEWART, Monika KUFFER

The census data provide a general indication which barangays are on average more effected by poverty

relevant issues, while spatial variation within them is hidden. Using the employed set of indicators pockets of

poverty could be detected leaving out poverty which is not spatially clustered.

5 CONCLUSIONS

The main thrust of this study was to develop an approach for mapping spatial indicators of poverty at a

disaggregated level to make spatial heterogeneity within administrative unit visible. To support such an

analysis, detailed spatial information is needed on poverty. A set of indicators was developed with the

assistance of local experts and stakeholders reflecting poverty aspects of morphology, location and living

conditions. Indicators related to morphology and location were used to find pockets of poverty while

indicators on living conditions gave insights into the diversity of these areas.

The analysis of morphology was operationalised by indicators on size of dwellings, density characterised by

the extreme clustering of roof coverage, lack of proper road network characterised by irregular layout of

settlements and non-visibility of access roads, poor structural quality of buildings regarding roof material.

These indicators were analysed using a very-high resolution satellite image. The criteria used on the image

were based on the interpretation elements of colour, texture, pattern, shape and size. The other relevant

datasets to assist in the identification were the point locations of the Homeowners Associations, dry hydrants

and flood prone areas. The integrated datasets could provide detailed spatial data for identifying poverty

pockets.

The results of data analysis at such a disaggregated level have shown where the critical poverty areas are

located and identified critical issues of sanitation and access to water supply affecting these areas. The

critical issues identified showed clearly that poverty is multidimensional in nature. The analysis could

indicate the barangays with the highest percentage of poverty pockets (e.g. Ermita with 27% of poverty

pockets) but also localise specific pockets of poverty with very critical living conditions. Such critical living

condition could be e.g. localised in two areas within Basak San Nicholas and Mambaling which had the

highest percentages of 43% and 42% respectively for households without water and toilet, were the main

issue is water supply and to a lesser extent sanitation.

Analysis based on the data integration and correlation of the results at the disaggregated and aggregated

levels gave credence to the fact that spatial variation can be hidden when data is aggregated. The results of

the data analysis at the aggregated level have proven the robustness of the selected indicators in analysing

poverty at the disaggregated level when comparing the percentages at the two levels. However, from the

analysis it was apparent how aggregation can mask variation of poverty within the barangays since, poverty

can be not localised and is averaged within one barangay. The specific areas (pockets of poverty) and their

spatial extent were clearly visible at the disaggregated level. The combination of both, analysing poverty

relevant indicators on administrative sub-city units in combination with localising the specific pockets of

poverty has good potential to assist in better targeting interventions.

6 REFERENCES

Baud, I.S.A., Sridharan, N. & K. Pfeffer: Mapping urban poverty for local governance in an Indian mega-city; the case of Delhi.

Working Paper 1 for IDPAD project New Forms of Urban Governance, presented at the New Delhi Conference of the

Project of New Forms of Urban Governance, June 2006. Davis, B.: Choosing a Method for Poverty mapping. Food and Agriculture

Organization of the United Nations. Rome, 2003. Etemadi, F.: Towards Inclusive Urban Governance in Cebu. Working

paper 25 on Urban Governance, Partnership and Poverty.

University of Birmingham, UK, 2001. Hall, G. B., Malcolm, N. W., & J. Pinowar: Integrating Remote Sensing and GIS to detect

Pockets of Urban Poverty: The case of

Rosario, Argentina. Transaction in GIS, Vol. 5, No. 3, pp. 235 253, 2001. Henniger, N. & M. Snel: Where are the Poor?

Experiences with the Development and use of Poverty maps. World resources

Institute, Washington, D.C., USA and UNEP/GRID. Arendal, Norway, 2002. Lemma, T., Sliuzas, R. & M. Kuffer: A participatory

approach to monitoring slum conditions: an example from Ethiopia. IIED:

Mapping for change: practice, technologies and communications. Participatory learning and action, No. 54, pp. 58-66.

London, UK, 2006. Martinez-Martin, J.: Monitoring intra urban inequalities with GIS-based indicators: with a case study in

Rosario, Argentina.

Utrecht University, Utrecht, ITC, Enschede, Netherlands, 2005. Moser, C. O.: The Asset Vulnerability Framework: Reassessing

Urban Poverty Reduction Strategies. World Development, Vol. 26,

No. 1, pp l-19, 1998. Turkstra, J. and Raithelhuber, M.: Urban Slum monitoring. ESRI Conference Paper No. 1667, available online:

http://gis.esri.com/library/userconf/proc04/docs/pap1667.pdf, 2004. United Nations Department of Economic and Social Affairs

(UN/DESA): The Millennium Development Goals Report, available

CEIT ALANOVA

857

Central European Institute of Technology,

Dept. for Urbanism, Transport,

Environment & Information Society

Where are the poor? A disaggregation approach of mapping urban poverty

online: http://mdgs.un.org/unsd/mdg/Resources/Static/Products/Progress2006/MDGReport2006.pdf. New York, 2006. Zeller, M.,

Sharma, M, Henry, C. & C. Lapenu: An Operational Method for assessing the poverty Outreach Performance of

Development Policies and Projects: Results of Case Studies in Africa, Asia and Latin America. World Development

Vol. 34, No.3; pp. 446-464, 2006.

REAL CORP 007: To Plan Is Not Enough: Strategies, Plans, Concepts, Projects

858 and their successful implementation in Urban, Regional and Real Estate Development

Contact this candidate