Geo-ethnography: Coupling Geographic Information Geo-Enthnography.pdf project (henceforth referred

Download Geo-ethnography: Coupling Geographic Information Geo-Enthnography.pdf project (henceforth referred

Post on 05-Jun-2018

216 views

Category:

Documents

4 download

TRANSCRIPT

  • JOURNAL ARTICLE 05-09

    STEPHEN A. MATTHEWS JAMES E. DETWILER LINDA M. BURTON

    Geo-ethnography: Coupling Geographic Information Analysis

    Techniques with Ethnographic Methods in Urban Research

    The article is available below and at utpjournals.metapress.com, University of Toronto Press, 40(4): 75-90.

    http://utpjournals.metapress.com/content/22881450w061r664/fulltext.pdf

  • Geo-ethnography: Coupling GeographicInformation Analysis Techniques with

    Ethnographic Methods in Urban Research

    Stephen A. MatthewsDepartment of Sociology / The Pennsylvania State University /

    University Park / PA / USA

    James E. DetwilerDepartment of Geography and World Campus Certificate Program in GIS / The Pennsylvania State

    University / University Park / PA / USA

    Linda M. BurtonCenter for Human Development and Family Research in Diverse Contexts / The Pennsylvania State

    University / University Park / PA / USA

    Abstract

    This research article focuses on the coupling of geographic information system (GIS) technologies with ethnographicdata, an approach we refer to as geo-ethnography. The data used here were gathered in an ongoing, multi-site study oflow-income families and their children. Throughout our work, the goals have been to think creatively about how GIScan be used in welfare research, to stretch the technology, and to revise the methodologies we currently use. Wespecifically discuss the ways in which the ethnographic data on families and neighbourhoods have been integrated withina GIS and how these two methods, alone and in combination, help situate families actions and experiences in timeand space and enhance data analysis and interpretation. More specifically, we focus on conceptual and methodologicalissues we have faced in the process of this integration and on practical strategies for combining qualitative andquantitative research.

    Keywords: Geo-ethnography, low-income populations, welfare, family and neighbourhood research

    Resume

    Larticle porte sur lassociation des technologies liees aux systemes dinformation geographique (SIG) et des donneesethnographiques, une approche que les auteurs appellent geo-ethnographie. Les donnees de larticle ont ete recueilliesdans le cadre dune etude multicentrique encore en cours sur les familles a faible revenu et leurs enfants. Durant letude,les auteurs ont tente de trouver de nouvelles manieres demployer les SIG dans les recherches sur laide sociale, afin derepousser les limites de la technologie et de revoir les methodes actuelles. Plus particulierement, ils ont examine lesmoyens par lesquels les donnees ethnographiques sur les familles et leur voisinage ont ete integrees dans le SIG etcomment ces deux methodes (donnees ethnographiques et SIG), seules ou en association, aident a situer dans le tempset lespace les actions et les experiences des familles et ameliorent lanalyse et linterpretation des donnees. En outre,ils se sont concentres sur les problemes conceptuels et methodologiques decouverts dans le processus de cette integrationet sur les strategies pratiques visant a combiner les donnees qualitatives et les donnees quantitatives.

    cartographica (volume 40, issue 4) 75

  • Mots cles: geo-ethnographie, populations a faible revenu, aide sociale, recherche sur les familles et leur voisinage

    Introduction

    The effect of recent welfare reform on low-income womenand children is the substantive focus of the Welfare,Children and Families Study, a complex, multi-siteproject (henceforth referred to as the Welfare Project).This article focuses on how geospatial data andgeographic information systems (GIS) have beenvariously adopted in the ethnographic component of theWelfare Project. It is important to note that GIS was notpart of the original design of the Welfare Project, andthus, throughout this article, we identify some of thechallenges and opportunities faced in promoting GISwithin multi-site ethnography and describe some of theways we have used GIS to facilitate distance-basedethnography. Our goals for the Welfare Project havebeen to think creatively about how GIS can be used tostudy the geographies of families, to stretch the GIStechnology, and to revisit and challenge importantconceptual and methodological issues regarding studiesof families and definitions of neighbourhood or context.

    An advantage of the mapping and data visualizationcapabilities of a GIS is that the system can handle datacollected on multiple spatial scales. Moreover, the dataobjects need not be restricted to numerical data, as a GIScan handle hot links to a variety of audio, video, image,and text files. In this way an ethnographic researcher cancombine, using a geographic framework, different formsof data (narrative text, photographs, audio and videofiles) and data layers in previously impossible ways. Inthe context of our work we use GIS to explore and betterunderstand the lives of the low-income families beingstudied. Our work suggests that combining GIS andethnography, an approach we call geo-ethnography,has many advantages for ethnographic research on low-income families. We caution, however, that GIS and theresulting maps and/or derived new information are notenough by themselves. In the context of our study, we stillneed ethnographic data to understand the child andfamily factors, the cultural meanings of place(s), and thepolitical and sociocultural influences on the day-to-daylives of low-income families. We still need ethnography tohelp us piece together how many of the contextual datasets that can be spatially joined are relevant to low-incomefamilies and low-income communities.

    We begin this article with description of the WelfareProject, with an emphasis on the main ethnographiccomponents. Next we describe the ethnographic dataproducts generated by the Welfare Project, therebypreparing the ground for a description of how GIS wasintroduced to the project and the examples, appearinglater in the paper, of how GIS is being used to facilitatemulti-site ethnographies of families and neighbourhoods.

    Lastly, we reflect on practical strategies for combiningqualitative and quantitative research.

    An Overview of the Welfare, Childrenand Families Study

    In 1996, the Personal Responsibility and WorkOpportunity Reconciliation Act (PRWORA) changedthe welfare landscape in the United States. PRWORAand the accompanying state legislation have been describedas the greatest single shift in social policy for low-incomefamilies since the passage of the Social Security Act of 1935;in President Clintons words, they represented the endingof welfare as we know it.1 The Welfare Project is anintensive study in Boston, Chicago, and San Antoniodesigned to evaluate the consequences of welfare reform forthe well-being of children and families and to follow thesefamilies as welfare reform evolves (see Welfare, Childrenand Families 2002 for additional resources).

    The Welfare Project includes a focus on children andchild development, includes a disability component,incorporates qualitative data from an extensive ongoingfamily and neighbourhood ethnography, includes bothwelfare recipients and non-recipients, is based in threedifferent geographical contexts, and is longitudinal indesign (see Winston and others 1999 for specific details).The project design has three main components: a survey,an embedded development study (EDS), and an ethnog-raphy (see Figure 1). All GIS activities in the WelfareProject are linked directly to the ethnography but not tothe survey or EDS components.

    The survey component includes interviews with some 2400households with children in low-income neighbourhoodsin the three selected cities. Approximately 40% of thefamilies were receiving cash welfare payments at the timeof the first interview in 1999. Each household includeda child aged 04 or 1014 at the time of the interview;that child and his or her primary caregiver are the focalparticipants. A second and third survey were completedin 2000/2001 and 2003/2004. The embedded developmentstudy (EDS) component incorporates the videotaping andcoding of caregiverchild interactions, time-study diaries,and observations of childcare settings. The approximately700 young children in the EDS are drawn from a sub-sample of families participating in the survey.

    The ethnography component, the primary focus of the GISactivities on the project, is ongoing in 25 neighbourhoodsacross the three cities; the ethnography team is tracking256 African American, Latino, and non-Hispanic whitefamilies residing in, or living in close proximity to, theseneighbourhoods. The ethnography includes three mainparts: family ethnography, disability ethnography, and

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    76 cartographica (volume 40, issue 4)

  • neighbourhood ethnography. The 211 participants in thefamily ethnography were recruited through formal andinformal sources such as community groups and serviceagencies as well as through introductions from otherstudy participants. Most families included at least onechild aged 24, and all were low income, with approxi-mately half receiving welfare at the time of recruitment.Forty-five families included a child under age eight witha disability (the disability ethnography). Data collectionfor the family and disability ethnographies includedparticipant observation and taped, in-depth interviewswith families over a period of 18 months (completedin 2002/2003), continuing with follow-up interviewsonce every six months through 2004. The neighbourhoodethnography, which was designed to help us betterunderstand the role of institutional resources and socialnetworks, typically involved collecting data via key-informant interviews, attending local neighbourhoodmeetings, monitoring local newspapers, and descriptiveaccounts of neighbourhoods via walkthroughs andinterviews at community-based organizations, especiallythose involved in child services and health care.

    The specific aim of the ethnographic component (family,disability, and neighbourhood ethnographies) is to betterunderstand how welfare recipients experience changes inwelfare regulations and how the decisions and behavioursof low-income families are influenced by the welfare

    system. The primary research questions arising from theethnography focus on the transition from welfare to workand back, making it financially, health and health care,parenting, social networks, caring for the disabled, andneighbourhood effects on all of the above.

    Structured Discovery, Ethnographic Data Journeys,and Ethnographic Data Products

    Within the Welfare Project, we characterize the ethnog-raphy as being one of structured discovery, an approachthat focuses on primary research topics while building insufficient flexibility to capture emergent themes andunanticipated information (Burton and others 2001).That is, many of the data on families take the form offield notes describing naturalistic, loosely structuredencounters in which field ethnographers observebehaviours and ask questions about primary researchthemes based around modules developed by seniorresearchers on the project. These modules cover discus-sions of the typical day, welfare reform experiences,household/family structure/power, residential mobilityand housing, education, family routines and rituals,family and social support networks, childcare, child andmaternal health, and child development, as well asdiscussions of neighbourhood characteristics.

    Thus, the approach to fieldwork revolves around struc-tured opportunities for discovery. For example, in the

    Ethnography Survey

    SurveyEthnographyFamily

    Ethnography

    Disability

    Ethnography

    Neighbourhood

    Ethnography

    EmbeddedDevelopmental

    Study(EDS)

    2,400 Families

    Estimated 80% of families with current incomes below poverty; 20% between the poverty line and 200% of the poverty line

    Estimated 40% on TANF

    Embedded Developmental

    Study

    Subsample of the families in the survey

    700 young children age 2 to 4 and their care givers

    Home observations and child care observations every 18 months

    215 African-American, Mexican-American, Puerto Rican, and Non-Hispanic White Families interviewed 18 to 24 months

    Follow-up interviews will take place every 6 months until completion of study

    TANF and Non-TANF

    45 Families with a child under 8 years old with a disability

    Key informant interviews, attending neighborhood meetings, neighborhood walkthroughs

    Figure 1. The design of Welfare, Children, and Families: A Three City Study.

    Geo-ethnography: Coupling Geographic Information Analysis Techniques with Ethnographic Methods in Urban Research

    cartographica (volume 40, issue 4) 77

  • context of health-related issues, ethnographic data collec-tion includes multiple encounters with families and greatlyfacilitates an exploration of multiple health events or issuesfor any family member. Such health events or issues mightinclude routine events such as health clinic visits andexperience of common ailments, issues such as domesticviolence and substance abuse, and discussions of concernsover local environmental risks such as industrial pollutantsor crime. By way of contrast, the survey component of theWelfare Project includes an array of questions andinstruments for gathering valuable data on the primarycaregiver, such as whether or not she has health insurancethrough her employer and whether or not her children arecovered, but these questions are for the most part closed-ended in format, precluding the ability to explore withrespondents additional questions on health-related topics.

    ETHNOGRAPHIC DATA JOURNEYS

    Ethnographic data are gathered from families andneighbourhoods in Boston, Chicago, and San Antonioand then transferred to two additional sites: the Universityof North Carolina (UNC) at Chapel Hill and thePennsylvania State University (PSU). Data on familiesthat include a child with a disability, a special feature ofthe Welfare Project, are transferred to UNC, and allethnographic data are delivered to the coordinating siteor depository at PSU. To gather, organize, analyse, andinterpret the data for this project requires a large team.The ethnographic team consists of more than 80 seniorethnographers and research scientists, family and special-ized disability ethnographers, neighbourhood ethno-graphers, coders, and programmers across five sites.All ethnographic data (tapes, transcripts, field notes,documents) from the three cities are archived at thecoordinating site, where coders read field notes ontheir assigned families and recode data in NUD*IST(Non-numerical Unstructured Data Indexing Searchingand Theorizing; see Gahan and Hannibal 1998). Eachmeeting between an ethnographer and a family (typicallythe primary caregiver) is written up, coded, and recordedin separate data files, with all text coded into categoriesor nodes. Each node is a three-letter identifier used tohighlight text, typically a paragraph within a field note,where reference is made to a specific research topic ortheme. In the family ethnography we have at least 12groupings of nodes, each containing up to a dozenspecific sub-themes. For example, 11 nodes are groupedunder Family Codes, including FEC, which specificallyrefers to family composition; FRL, referring to familyrelations; and FRT, referring to family routines.Other groupings of nodes include Actors (i.e., key subjectsin the family/household), Health and Nutrition,Home, Neighbourhood (discussed in more detail below),Perception, Relationships, Welfare/Services, Work andEducation, and Other, as well as groupings of nodes for

    ethnographers reactions and other logistical matters.Each paragraph or section of text within a field notecan be tagged with multiple nodes. Within NUD*IST,all field note and document files are searchable by text(words and phrases) or nodes. For example, to retrieveinformation on transportation and access to health care,one might request a search on the nodes TRP and HTH.

    ETHNOGRAPHIC DATA PRODUCTS

    The team at the coordinating site not only codes andrecodes data extracted from transcripts and field notesinto NUD*IST but also generates a number of specializeddatabases on each of the families and neighbourhoods.These specialized databases include family profiles, a life-at-a-glance calendar, and demographic tables. Familyprofiles are evolving documents rewritten after the receiptof each field note. Profiles are structured around primaryand discovered research topics and include selectedinformation extracted from field notes (e.g., illustrativequotations around key research themes such as transi-tioning from welfare to work, parenting, social networks,access to and use of health care). The life-at-a-glancecalendars are one-page tables in which columns representspecific months of a year and the rows represent keyevents in the lives of a family, recorded as either events orspells (e.g., birth, death, marriage, cohabitation, residen-tial move, on and off welfare, sanctioning, illnesses,employment and unemployment). These calendars area visual tool for conveying salient events in the lives ofthe families and facilitate the identification of temporalclusters or contemporaneous occurrence of events orexperiences. The demographic tables describe the familiesat the time of their recruitment to the study and includeinformation on welfare experiences prior to recruitment,TANF (Temporary Aid for Needy Families) status,disability status, marital status, family size, age of familymembers, place/country of birth, and so on. In additionto these specialized data sets, other data files includespreadsheet data listings of all neighbourhood resourcesused by families and children, as well as education andresidential history files (see examples below). Thus, datagathered as part of the Welfare Project can be found ina variety of text file formats (Word, NUD*IST) andspreadsheets (Excel, DBF), as well as analog audio-visualformats (taped interviews, photographs).

    These various ethnographic data products not onlyinclude rich data pertaining to our primary researchquestions, they also include information on specificlocations and areas within and beyond the three cities.For example, during an interview with a family, referencemay be made to a specific location (e.g., a daycare facility,Head Start program, place of work, welfare agency), toresources in the community (e.g., a library, church, orcommunity group), or to the locations of family membersor members of a social network. Similarly, interviews

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    78 cartographica (volume 40, issue 4)

  • might focus on individual or family routines such as thejourney to work, shopping trips, social events, or regularvisits to a health centre. Thus, in the family and disabilityfield notes we frequently find references to specific placesand journeys. Within the NUD*IST databases there arenodes for an array of neighbourhood-related issues suchas the characteristics of the neighbourhood (NCH),available resources within the neighbourhood (NRS),and perceptions of neighbourhood (NPR). Similarly,a catch-all NUD*IST node, GIS, refers to geographicinformation on any location such as an address, crossstreet, boundary street, or place name. Not surprisinglydata coders have added GIS nodes to many of theneighbourhood, family, and disability field notes.

    Many of the geospatial data described above might havebeen collected in any ethnographic project similar to ours,if by accident rather than design. However, in ourscanning of ethnographic research we have found fewinstances of the explicit use of primary or even secondarygeospatial data to visualize the lives of individuals. Thisshould not be taken as indicating that the coupling ofmapping and ethnography has not been discussed (for auseful primer, see Cromley 1999). Similarly, the applica-tion of time geography and space-time activity studies,though not necessarily drawing on ethnographic data,has seen a rejuvenation of interest in the past few years,led by geographers such as Mei-Po Kwan (see, e.g., herprogress report on time, information technologies andthe geographies of everyday life; Kwan 2002a, 2002b).

    Introducing GIS to Ethnographerson the Welfare Project

    The coordinating site pursued multiple strategies tointroduce GIS concepts and methods to ethnographers.Throughout the project we have worked with one or moresites to create new and integrate existing geospatialdatabases within a GIS framework. The intent was bothto demonstrate to the sites what we could do with the datathey collect and to encourage the collection of geographicidentifiers, however crude, on the location of familyactivities and neighbourhood resources within each city.

    Prior to this study we (authors Burton and Matthews)collaborated on projects where information derived fromethnographic studies had been presented in map form(e.g., Burton and Graham 1998). Since these early foraysinto combining data sources and method, our goal hasbeen to think creatively about how GIS can be used inethnographic research on low-income families, to stretchthe technology, and to revise the methodologies we use.Shortly after the Welfare Project received its primaryfunding and the ethnographic teams in each of the threecities were gearing up for pilot-phase work, we weredetermined to introduce our colleagues to the potentialbenefits of collecting geospatial data within the

    ethnographic component. The first example using geos-patial data that we shared with the ethnographic teammapped out hypothetical journeys that reflected thetypical daily routines of a young mother juggling work,childcare/schooling, and other family obligations (seeFigure 2). It is important to note that while this map isbased on fictitious data, the actual journey types (multi-mode journeys consuming considerable time and traver-sing relatively long distances) are common amongWelfare Project study families.

    Prior to these early team meetings, few of our colleagueshad seen mapped representations of data gatheredthrough ethnographic fieldwork or based on space-timeactivity data. They were unfamiliar with Hagerstrands(1970) time geography and space-time activity pathways(Rose 1993; Golledge and Stimson 1997) and geographicresearch that has mapped out daily routines and/or keyanchor institutions in the lives of, and over the life courseof, individuals or groups (see Rowe and Wolch 1990;Gilbert 1997, 1998; Laws 1997; Kwan 1999, 2000, 2002b;McLafferty 2002).

    The reception of the hypothetical map and the introduc-tion to geographic information systems in general weredecidedly mixed. The value of the ethnography is theprolonged contact with families and the resulting richdescriptive data that provide in-depth and nuancedunderstandings of what it is like to be low-income andminority in an era of welfare transition. In contrast, whenGIS was first introduced to members of the ethnographicteam as a methodology to use in tandem with ethno-graphic analysis, there was reluctance on the part of someresearchers. Some were not convinced that the time spentcollecting the necessary geographic information wasworth whatever additional analysis GIS could provide.Moreover, they were not convinced that GIS wouldcontribute significantly to interpretations that could bemade from textual data alone or that GIS would offera better portrayal of families lives than a coherentethnographic narrative. There was some concern thatcomplex stories would be simplified and flattened outin two-dimensional models and that, if taken out ofthe context, without reflecting change over time, thesemodels could distort the analysis or be misread (Skinner,Matthews, and Burton 2005). In short, convincing theethnographic team of the value of collecting geospatialdata on families through the ethnography was not easy.

    Although the maps were not always well received, thisfact was partially offset by positive curiosity about thegeospatial data used to help construct the maps. That is,our colleagues were interested in knowing more about thegeospatial data available on streets and transportationnetworks and, in other examples not reproduced here,the various census geographies and other areal unitboundaries such as police districts, as well as a varietyof point feature databases such as the locations and

    Geo-ethnography: Coupling Geographic Information Analysis Techniques with Ethnographic Methods in Urban Research

    cartographica (volume 40, issue 4) 79

  • attributes of Women, Infants and Children (WIC)program clinics. They came to appreciate the contextualgeographic databases, specifically the neighbourhoodand city profiles, that a GIS could be used to create.It no doubt helped that few others on the project werefamiliar with demographic, urban, and infrastructure datathat could easily be integrated within a GIS.

    DEVELOPING CONTEXTUAL DATABASES

    In the early stages of the Welfare Project, the GIS activitiesfocused on developing a comprehensive geospatialdatabase for the three cities and, specifically, for theneighbourhoods we were studying. At its most straight-forward, this process consisted of gathering as manyrelevant and timely secondary data as were availableeither free or at modest cost. In effect, we were buildinga geospatial database in anticipation of requests forbasic locator maps and maps that might display keyneighbourhood resources. That is, we were developingGIS projects (one for each city) that would enable oneof the GIS analysts on the project to construct a desiredmap upon request.

    The initial geospatial databases included street files (andpost-processing of these data to differentiate line segmentsinto freeways, major roads, roads, etc.); rail transportationsystems (routes and stop locations); land-use data(presence of water bodies, parks, etc.); an array of areaboundaries, including ZIP code and census geographies(e.g., block group, tract); and attribute informationfor these areas derived from the 1990 and 2000 census(e.g., race/ethnic breakdown, poverty). On this base weadded data from a variety of official sources and otherdata providers. Our goal was to move beyond a limitedset of census-derived variables and seek out practicaland innovative uses of alternative data (e.g., on crime,health, land use, and transportation) that describe andcapture dimensions and characteristics of neighbour-hoods not previously contemplated (see Coulton 1997;Sampson, Moronoff, and Earls 1999; Burton and others2000). Our geospatial database for each city includesaerial photos from the mid-1990s and publicly availabletax, property, and land-parcel data. The latter have provedinvaluable in generating up-to-date land-use profiles ofour study neighbourhoods and have greatly facilitatedaddress-matching tasks. To these data, depending on city,

    Hypothetical journey towork for a young

    mother in San Antonio(time 2 hours one way)

    0 1 2 3 4 miles

    N

    1. Residence (leave home around 6:30 am to walk to the day care centre with two children)

    2. Day Care Centre (arrive at day care at 6:45 am and drop off youngest child, then leave at 6:50 am)

    3. Elementary School (arrive at 7:00 am and drop off older child, then leave at 7:05 am to walk to the bus stop)

    4. Bus stop (arrive at bus stop 7:12 am to catch first bus around 7:17 am)

    5. Bus depot (arrive around 7:44 am wait to catch second bus around 7:54 am)

    6. Bus stop (arrive at bus stop near place of work around 8:22 am then begin short walk to work)

    7. Place of Work (arrive at work 8:30 am)

    The return journey home can be morecomplicated. For example, the mother may shopin area (9) near the school and day care centrebut which requires a different bus route-betweenthe bus depot (5) and bus stop (8). After groceryshopping the mother walks to the elementaryschool and day care to collect her children, and then continues home.

    StreetRailroad

    Figure 2. A hypothetical journey.

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    80 cartographica (volume 40, issue 4)

  • we have added spatially and temporally disaggregatedcrime data; integrated data on the locations of family,group, daycare and after-school care centres and facilities,including attribute data on size, years of operation,and age ranges served; and data on all school locationsand attribute information on age/grade, sex, and race/ethnicity profiles within schools.

    Finally, from the family, disability, and neighbourhoodethnographies (and pilot studies in each of the threecities), we added geospatial data including point featuressuch as family home addresses at recruitment and thelocations of recruitment sites (e.g., Head Start, commu-nity groups), as well as some area features such asperceived neighbourhood boundaries as defined andderived from key informant interviews within theneighbourhood ethnography. We also created study-areaboundary files based on hard-copy maps of the censusblocks used in the Welfare Projects survey componentsampling frame. These secondary data sources wereacquired, compiled, processed, and collated by the GISteam at the coordinating site.

    TOWARD A VIRTUAL ETHNOGRAPHY

    Experienced GIS users will realize that simply havingconstructed a geospatial database is not the same asmaking the database easily usable or accessible to noviceusers, such as many members of our own ethnographicteam (i.e., senior and field ethnographers in the threecities and the data coders at the coordinating site). Tocircumvent some of the problems of access and utilizationof these data, we began to explore ways in which graphicaluser interface tools could be embedded in our GIS projectto help novice users navigate both the software andthe extensive geospatial data sets being constructed fromsecondary sources. This work was extended furtherwhen we began to work directly with primary data (i.e.,geo-coded ethnographic materials, such as family profilesand neighbourhood walkthroughs).

    The interface, known as the Welfare Extension (writtenin Visual Basic for use in ESRIs ArcGIS software byco-author Detwiler), allows the user to visualize andexplore contextual geospatial data and their associatedattributes in relation to family locations and neighbour-hood boundaries (Figure 3). The user is presented with anopening map view for the city of interest containing threelayers: a layer demarcating the ethnographic studyneighbourhood boundaries and two index layers thatallow for efficient retrieval of orthophotographs (aerialphotographs) and property-level data respectively. Afterthe user selects the neighbourhood(s) of interest, the mapview is automatically re-scaled and centred. The user canthen select from an array of data available for theneighbourhood(s). Current options allow the user toretrieve and map neighbourhood ethnographic data

    derived from walkthroughs and key informant interviewsor to map family locations as well as selected census,property/land-use, school, crime, and street file data(these data options are continually being extended toinclude both more diverse types of geographic data andvarious family ethnographic data products such as thedemographic and family profiles).

    Examples of Coupling GIS and Ethnographic Data

    To date, we have combined GIS and ethnographic datain a variety of different ways, many of them based on theretrieval of relevant data through the use of the WelfareExtension. The first three examples that follow are takenfrom the different components of the ethnography(family, disability, and neighbourhood components),while the fourth draws on the integration of geospatialdata from both family and neighbourhood ethnographies.

    FAMILY ETHNOGRAPHY

    As described above, the family ethnography includes anextensive set of field notes on multiple topics. In recentyears we have created maps and images for conferencepresentations and reports using data from original fieldnotes, family profiles, and the life-at-a-glance calendar(Burton and others 2000; Burton, Hurt, and others 2001;Matthews, Burton, and Hurt 2002; Skinner and others2005). Thus far we have used GIS to help describe afamilys local contexts (neighbourhood resources suchas various community services and risks such as crime);to explore the spatial and temporal constraints faced bylow-income minority families (e.g., through the mappingof home, daycare, school, and employment locations);and to document how families use the neighbourhoodsand cities in which they reside (e.g., social networks,shopping activities).

    Because of their detailed nature, both ethnographic andgeospatial data can easily be used to identify individualsand neighbourhoods in our study. In recent years, GISand concerns over privacy have received much criticalattention (Pickles 1995; Curry 1997, 1998). FollowingMark Monmonier (2002, 1), we note that the maps onelooks at are less important than the spatial data systemsthat store and integrate facts about where we live andwork. For obvious reasons (data restrictions, confidenti-ality, and privacy concerns), we are not permitted toproduce maps that could easily identify individuals.In dealing with the maps one looks at, we employ severalmapping strategies to avoid privacy violations (Cromleyand McLafferty 2002). Such strategies include, but are notlimited to, turning off geospatial data layers (such as roadnetwork data) so as to avoid providing location clues;displaying data on small-scale maps; aggregating data; andusing masking techniques. Regarding spatial data systems,the data from the project used in a GIS environment

    Geo-ethnography: Coupling Geographic Information Analysis Techniques with Ethnographic Methods in Urban Research

    cartographica (volume 40, issue 4) 81

  • are housed in restricted data directories within aKarlbridge firewallprotected computer network at thehost institution, and access is limited to key projectpersonnel and GIS programmers, all of whom arerequired to have completed the University InstitutionalReview Boards human subjects training.

    The family ethnography includes a module or topicalinterview focusing on residential mobility. In the examplebelow, the ethnographer recreated from original fieldnotes a listing of all known residential locations and thedates when moves between locations occurred. Thisinformation was entered into a spreadsheet and thenaddress-matched. Figure 4 recreates the residential historyof a young mother, referred to as Maria. Maria has moved13 times during her life, including five moves while livingwith her parents as a child. She left home in 1990 and hassince made eight moves, mostly within a relatively smallarea approximately four miles west of Chicagos Loop.Many of the location changes were of short duration, withmoves in the early 1990s frequently involving movingin with another family member. One of the challenges ofmapping residential histories gathered in ethnography isnot to infer too much from the resultant pattern withoutan adequate cross-reference and reading of field notesregarding other significant events or periods in the lives

    of our families. That said, mapping the data has tangiblebenefits, such as revealing interesting patterns that mayhave been missed by the family ethnographer, and, at aminimum, may help generate new questions regardingresidential stability and mobility among low-incomefamilies.

    DISABILITY ETHNOGRAPHY

    Families in the disability ethnography face uniquechallenges, juggling work, welfare requirements, andfamily needs. To explore time and space constraints,the disability ethnography component experimented withthe collection of time diaries for a period of one month(November 2000). The intent was to capture location andduration information on all meetings and/or appoint-ments, as well as the journeys undertaken by the primarycaregiver of a child with disabilities. The data werecollated, entered into a spreadsheet, and address-matched.Maps were generated in part to illustrate the complexityof disability management (Skinner and others 2005).

    In the example below, the primary caregiver is a mother,aged 30, who has a seizure disorder and partial paralysisdue to a stroke at age 5. At the time of this time-diarystudy, her family included a 14-month-old son (focal

    Step 1: Select City View

    Step 2: Select Neighbourhood(s)Step 3: Select Data type

    (and drive for data access clearance)

    (from the list described at right)

    Ethnographic DataSelecting this item generates a listof geocoded neighbourhoodwalkthroughs and key informantinterviews. These data files areuniquely identified based on theneighbourhood ID and date offield note (e.g. B0l-030700). Thefield notes include geocodes tospecify an address, cross-streetsor journey segment and thus areloaded into the project as a newtheme. Using hot link feature thefield note can be viewed bymoving the cursor over the point(address)or linear (journey or

    Census DataSelecting this item generates a listof nineteen census variablesincluding measures that focus onrace, ethnicity, place of birth, yearof entry, family type, educationalattainment, income and poverty.The selection of a variablegenerates a new layer thatincludes either a choropleth mapora pie-chart (proportional in sizeto the population universe ofinterest). These new layers are areasonable representation of thedata with out expecting noviceusers to design their own mapbased on legend editor selections.

    Crime DataSelecting this item generates a listof property and violent crimevariables. Note that crime datavary widely by city (spatial andtemporal resolution). In the caseof San Antonio specific propertyand violent crime data are actualpoint events that can be selectedby date of occurrence (e.g., allburglaries in Jan 2000). The userselects crime type and defines thetemporal period of interest. Thisgenerates a theme meeting theselection criterion.

    Property DataSelecting this item generates a list of property attributes. Propertydata vary by city but share manycommon features such as majorland use categories (e.g.,residential, commercial,industrial, vacant, tax-exempt,etc.). Similarly, specific land usescan be selected such as religiousproperty, libraries, schools,hospitals, parks and so on. Thelatter selection generates a newshapefile that includes allproperties that meet the selectioncriteria that can be found insidethe neighbourhood and within amile of the neighbourhoodboundary.

    Family DataSelecting this item generates anew point theme that identifiesthe approximate home location ofthe families in the study at thetime of their recruitment. Onlyfamilies with the appropriateneighbourhood ID field aremapped. While actual addressinformation is known these dataitems have been stripped and theattached data file includes only afamily ID number.

    Road DataSelecting this item generates anew line theme that includes allstreets that can be found insidethe neighbourhood and within amile of the neighbourhoodboundary.

    Schools DataSelecting this item generates a listof attributes for public (schooltype, enrollment, racialcomposition, percent free lunchrecipients) and private schools(enrollment and racialcomposition). A new theme isadded to the view either as a pointsymbol (i.e., school type) or as apie-chart proportional in size tothe enrollment within the school.The selection generates a newshapefile that includes all schools(elementary, middle, high, specialand other) that meet the selectioncriteria that can be found insidethe neighbourhood and within amile of the neighbourhoodboundary.

    route) feature.

    Figure 3. The Welfare Project Extension.

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    82 cartographica (volume 40, issue 4)

  • child) with significant gastrointestinal problems, seizures,and allergies and a two-year-old daughter who has bentleg bones requiring a brace. The husband (father of thesetwo children) and two other children are present in thehome. Over the one-month period, the primary caregiverattended 29 meetings, of which 13 were held in her home(see Figure 5). The home-based meetings were typicallyfor counselling or physical therapy sessions for theyoungest child (twice a week on average), although theyalso included a Head Start home visit and a meeting withan attorney. The remaining 16 meetings were held at oneof 10 different locations across the city (two were closeto home and are obscured on this map by other data).These included six meetings at three different Head Startlocations; two visits to two different hospitals; threevisits to an early intervention centre; and visits to a healthclinic, a dentist, and a local housing authority office.The primary caregiver was accompanied by at least onechild for all but three meetings. Entering these datainto a calendar format reveals that all told, during thisone-month period, the primary caregiver needed to beat home for her sons physical therapy sessions orattended other mostly health-related meetings on17 separate days. As one might imagine, the time demandson this particular caregiver make holding a nine-to-fivejob or fulfilling welfare requirements somewhat difficult.While the family ethnographers are often familiar withthe complexity of the lives of the families they study,the visual representation, certainly in cases like the onedescribed above, often better reveal the challenges

    and constraints the families face and better groundthe ethnographic discussions with families aboutmanagement and coping strategies.

    NEIGHBOURHOOD ETHNOGRAPHY

    As mentioned earlier in the article, we have identifiedplaces and routes within neighbourhood ethnographyfield notes, focusing on field notes that coverneighbourhood walkthroughs. Early in the collection ofneighbourhood ethnographies we selected walkthroughfield notes to determine whether data gathered in thefield could be easily geo-coded and thus entered intoa GIS database. The field notes we selected proved usefulfor a number of many reasons, of which two areidentified here.

    First, it was evident that the neighbourhood walkthroughsprovided detailed text on a number of specific places(e.g., community resources and anchor institutions) thatcould be geo-coded. These data provide value added inbetter understanding neighbourhood resources and risk,especially when linked in with other secondary geospatialdata and even more so when linked to the geo-coded datagathered from the family and disability ethnographies.However, the neighbourhood walkthrough field notesdo create some problems. That is, neighbourhoodwalkthrough files inevitably refer to multiple places orroutes, requiring that we spent a great deal of timedividing up the field notes so that we could hot-link onlyrelevant text to the appropriate places or routes (indeed,

    0 1 2 3 4 miles

    ID YearStart

    YearEnd

    Number ofResidents

    Head of Household

    6

    7

    8

    9

    10

    11

    12

    13

    1990

    1991

    1991

    1992

    1992

    1994

    1994

    2000

    1991

    1991

    1992

    1992

    1994

    1994

    2000

    2000

    4

    3

    2

    3

    3

    3

    5

    10

    Boyfriends Mother

    Marias Grandmother

    Maria and Boyfriend

    Boyfriends Stepfather

    Maria and Boyfriend

    Maria and Boyfriend

    Maria and Husband

    Brother-in-Law

    .

    Figure 4. Recreating a residential history from ethnographic data.

    Geo-ethnography: Coupling Geographic Information Analysis Techniques with Ethnographic Methods in Urban Research

    cartographica (volume 40, issue 4) 83

  • we face a similar problem with data gathered via keyinformant interviews that is, it would be easy to geo-code the key informant field-note text file to the locationof the interview, but often the key informant refers tomany places within an interview, as well as locations inand around and even beyond the neighbourhood).It is important to note that we have faced challenges;specifically, we were unable to keep pace with the numberof files arriving from the study sites, to the degree thatgeo-coding neighbourhood data files is now performed onan ad hoc basis.

    Second, during the early stages of the neighbourhoodethnography we were able to demonstrate to theethnographic team an important data-verificationfunction of GIS. To demonstrate data verification, wedissected a neighbourhood ethnographers walkthroughfield notes and geo-coded specific text to places theethnographer observed and routes he or she walked.This allowed us to recreate or trace the ethnographersroute through the neighbourhood. This route was over-laid on the neighbourhood boundary as defined by theethnographic component. By mapping the route of theneighbourhood walkthrough, we can easily identifyinstances where the ethnographer is inside or outsidethe neighbourhood of interest. In an example taken fromthe early stages of the ethnography, an ethnographerwas observed to be outside the study area. Field notesreveal that the ethnographer is aware of considerable

    differences in physical structure and socio-economic andracial composition:

    I headed east again . . . the residential part of the neighbor-

    hood. Very different. Quiet. Working and middle class houses.

    Some were old. Several were recently renovated. I saw few

    people, but these were different races, Afro-American,

    white, and occasional Latino and apparently middle class

    professionals. (A neighbourhood ethnographers field

    note, Boston, 17 June 1999)

    The ethnographer continues on their journey,making additional references to the surroundings. Laterduring the walkthrough the ethnographer re-enters thestudy neighbourhood and immediately notices a change:

    I was heading back . . . . The neighborhood became less

    middle and more working class, leaning toward poor.

    Suddenly, I saw a few Latino faces again. I heard Latino

    music. The closer you get to [. . .] the greater the number

    of Latinos. (A neighbourhood ethnographers field note,

    Boston, 17 June 1999)

    While the ethnographer was aware that the characterof the neighbourhood and the people in it seemed to bechanging, the visualization of the descriptions hot-linked to known locations within a GIS could verifythat the ethnographer had indeed walked out of thestudy neighbourhood (in this particular instance,by almost half a mile) and identify when and where

    Monday Tuesday Wednesday Thursday Friday

    1a.Travel to/fromEarly Intervention Programb. Counseling for BC

    23Travel to/from

    EarlyInterventionProgram

    8a. Policy Committee meetingb. Progress Review for KC

    7Counseling for BC

    6Physical therapy for BC

    9Education Committee meeting

    10Cablevision

    15a. Parent meetingb. Renewal of

    Lease (BHA)

    14a. Dental visitb. Counseling for BCc. Travel to/from Orthopedic appt.

    13Physical Therapy for BC

    16Head Start 6-

    month check up

    17

    22a. Dental appt. for

    DCb. Interview for

    WelfareChildren andFamilies Project

    21a. Travel to/from

    Doctorb. Counseling for

    BCc. Travel

    to/from TownPolicy meeting

    20a. Dental appt. for

    KCb. Physical

    Therapy for BC

    23 24

    29Parent Group

    meeting

    28Counseling for

    BC

    27a. Appt. for DCb. Physical

    Therapy for BC

    30

    0 1 2 3 4 miles

    Calendar of events - one sample month

    MultipletripsSingletrip

    Figure 5. A map and a calendar representation of trips made and meetings attended during one month by a primary caregiverin a disability family.

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    84 cartographica (volume 40, issue 4)

  • the neighbourhood was re-entered. These techniques,while difficult to operationalize in near real time(without use of global positioning system [GPS] units),can nevertheless prove extremely useful in exploringdefinitions of geographically demarcated units and thefluidity of the boundaries of those units for, in our case,the low-income families of interest (see next example).

    NEIGHBOURHOOD AND FAMILY ETHNOGRAPHY

    We have integrated geo-coded neighbourhood and familyethnographic data to explore important conceptual issuesassociated with research on families and neighbourhoods.Perhaps the most important is the degree to whichneighbourhood context matters for families (see Brooks-Gunn, Duncan, and Aber 1997a, 1997b; Martinez 2000;Booth and Crouter 2001). Unfortunately, both spatialissues and the characteristics of neighbourhoods havebeen poorly conceptualized in family sociology anddemographic research. Indeed, reviews of the literatureon neighbourhood effects on child development, health,and well-being have been quite critical of past researchmethods and particularly of the data sets used (seeGephart 1997; Furstenberg and Hughes, 1997; Coulton1997; Leventhal and Brooks-Gunn 2000; Burton andJarrett 2000). As Furstenberg and Hughes report,

    the primary impediment to progress in understanding the

    effects of neighborhoods on children continues to be the lack

    of data sources that contain information on neighborhoods,

    families and children. (1997, 353)

    A key question is, How do we measure neighbourhoodor context? Much current urban demographic andquantitative research describes context or neighbourhoodas equivalent to a census tract, relying heavily on a limitedset of census variables, and rarely considers includingmeasures of neighbourhood resources or lack there of(e.g., retail, business, and health and social serviceenvironments, transportation infrastructure, and landuse) or neighbourhood risks (e.g., crime). In general,the tract-level studies offer weak and generally mixedevidence on the influence of neighbourhood effects (seereviews by Gephart 1997; Leventhal and Brooks-Gunn2000). A related problem is that the effects of the localenvironment on family or individual well-being presum-ably depend on how much the family or person is exposedto the neighbourhood where they live (Sastry and others,2002). Neighbourhood boundaries, as operationalized byresearchers, are likely to be less salient for individualswhose routine daily activities do not occur in theneighbourhood of residence. As Sastry and others note

    While there is a large body of research on issues such as

    journey to work, considerably less is known either about the

    overall spatial patterns of daily life or, more specifically, about

    the salience of neighbourhoods, whatever the definition, for

    individuals and families (2002, 3).

    Preliminary analysis of ethnographic data from Bostonfamilies reveals that census tracts do not capture manyactivities and/or locations relevant to a familys routineactivities (Matthews and others 2005). To paraphraseRonald Abler, John Adams, and Peter Gould (1971), someareal units are particularly sacred once they have beenestablished, even though they may later become seriousobstacles to solving contemporary problems. Figure 6areveals that three families who reside in the same censustract use and rely on resources and social networks acrosslarger areas than the immediate context as defined bythe census tract boundary (and even neighbourhoodsas defined by our own study not shown on the maps).Indeed, in the preliminary work based on data on morethan 1000 locations referred to in field notes from34 families, we find that most of their activities (usesof specific locations), indeed more than 90% of allgeo-coded activities, are outside the census tracts ofresidence. When these activities are broken down bydifferent domains of the families lives, we see, notsurprisingly, that those activities closest to home areassociated with food or grocery shopping, educationalactivities, recreation, health services, and other (i.e., non-health related) services. However, in no domain categoryis the percentage of anchor locations found in thehome and immediately adjacent tracts combined greaterthan 30% of the total, with food the most local ofall activities. Similarly, non-food shopping activities,work locations, and social support networks (perhapssomewhat surprisingly) are mostly found in distantlocales. Again, these findings are not altogether surprising,as most low-income families do not live near their eitherplaces of work or the larger shopping centres.

    As another example, Figure 6b shows the locations offamily and child resources in relation to a familys home.Again, this familys activity space, focusing on family andneighbourhood resources they refer to, is up to two mileseast of their residence and mostly outside their immediatecensus tract. The reason for the mismatch betweenresidence and use of resources, although not shownhere, is largely explained by the familys social networkand past residential history (i.e., a number of familymembers and friends reside in the area where anchorinstitutions and resources are concentrated, and themother has lived in and is familiar with this area).

    In the Welfare Project we use existing data sets, augmentthem with new and refined measures of spatial contextand structure of the urban environment, integrate thesedata with ethnographic data, and then analyse the datausing visualization/mapping techniques. Our applicationof GIS focuses on retrofitting data on contexts orneighbourhoods and on the integration of ethnographicdata on low-income families. This process results in a

    Geo-ethnography: Coupling Geographic Information Analysis Techniques with Ethnographic Methods in Urban Research

    cartographica (volume 40, issue 4) 85

  • better understanding of low-income families mobility,their use of and access to services, and an array of well-being-related issues. The interface or coupling tools wehave developed facilitate a more flexible viewing of ourstudy families and allow novice GIS users to visualizeand better understand the complexity of the lives oflow-income families and the strategies they adopt innavigating a fluid welfare reform landscape, therebyenhancing data analysis and interpretation. In ourwork to date we have explored the integration ofboth quantitative and qualitative data on families andneighbourhoods to help better understand different formsof spatial behaviour (Golledge and Stimson 1997). Wehave begun to explore different dimensions of mobilityand the use and knowledge of space by families, with anemphasis on the use and accessibility of services, housing,and employment; the presence and maintenance of socialnetworks; and the use of local and non-local opportunitiesor resources relevant to family well-being (Burton andothers 2000; Matthews and others 2002; and Matthews2005). In other words, we are exploring spatial andtemporal entrapment and the ability of families to jump

    scale, that is, to take advantage of non-local networks ofsocial and institutional support, stretching and redefiningtheir own complex family geographies (Aitken 1998).

    Some Challenges in Coupling GISand Ethnographic Data

    Combining methods makes clear that GIS has advantagesfor family ethnography (Skinner and others 2005). First,GIS has proved to be an effective tool by depicting, in asingle image, the intense effort it takes for primarycaregivers to create and maintain both work and familyresponsibilities and their social and service networksaimed at promoting family health and well-being aneffort that may surprise some, including policy makers(see Figures 2 and 5). The ethnographic description of afamilys efforts and experiences is still crucial, but forsome audiences, a GIS image may have an immediateimpact and provide a more powerful statement than anarrative account. Second, GIS is an effective tool inanalysis. Maps can suggest interpretations that we mightotherwise overlook. GIS can quickly identify families that

    32

    1 4

    5 6

    7

    89

    10

    11

    12

    These families used a varietyof resources outside of thelocal contextual area asdefined by census tract. Theexternal resources includedmedical services, workplacesand training locations, day careand after-school care facilities,shopping centres, religiousinstitutions and the homelocations of family and friends.

    a) b)

    Family resources used and referredto in family field notes

    123456

    Clinic/HospitalParkClinic/HospitalParkFamily/Womens ShelterAll-day Pre-school

    789101112

    Soup KitchenWIC/ Aid Food CouponsDaycare After-schoolLibraryESL/After-schoolPark

    Social Network

    0 1 2 3 4 miles

    CensusTract

    Figure 6. (a) Census tract as context? Spatial profiles of three families who reside in the same census tract. (b) A familys localanchor institutions and resources mentioned in ethnographic field notes.

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    86 cartographica (volume 40, issue 4)

  • are more isolated, have fewer supports, and access feweroutside resources and services. GIS can show how farfamilies have to go for different types of services thatenhance their childrens development and well-being.Thus, maps of families navigations may foster compar-isons not easily perceived through text alone. Third, GIScan also help us plot any significant changes in familiesconstructions or routines over time, such as identifyingchanges in the intensity and range of families use ofcommunity services and activities over time.

    In spite of these advantages of employing GIS technology,we caution that GIS and the resulting maps are notenough by themselves. For example, one of the challengesinherent in ongoing ethnographic research concernsanalysis, the salient questions being when and how toanalyse data. In many ways, analysis within ethnographyis an ongoing process, with each successive interview andfield note adding something new and prompting are-analysis. The GIS database evolves in parallel with theethnography, but the typical end product generatedfrom GIS is a static map, one that may quickly be outof date. Again, without reference to accompanyingethnographic field notes, the maps need not andprobably do not convey the whole story, and certainlythey cannot describe the dynamics of the individualor family story.

    The volume and diversity (text, audio, video, image) ofethnographic data is a key concern in any attempt tointegrate GIS and ethnography, raising many logisticalissues. Simply working with the data, verifying geospatialdata accuracy, and organizing data in ways that facilitatelinkage to a GIS project can be very time consuming,so much so that the Welfare Projects GIS team now dealswith map requests in reactive mode. That said, whena request is received to work with data on one or morespecific families (e.g., to produce a series of maps for aconference presentation or poster), our data managementsystem allows us to retrieve data products efficientlyfrom the secure project server, while the WelfareExtension facilitates the retrieval of important locatorand contextual data for specific neighbourhoods or forany area within the relevant city. Moreover, the GIS teamcan draw upon multiple team members (ethnographersfrom the relevant city as well as data coders at thecoordinating sites) to verify data and to ensure that weare representing the lives of individuals and familiesappropriately.

    In other respects, too, a GIS-focused approach is certainlynot a panacea. Handling primary and secondary geo-coded data can be a methodological minefield. In additionto concerns over access to theoretically relevant andtimely data, we also faced a number of challengesregarding the quality of the geospatial data collected bythe ethnographic team. At the beginning stages of theproject, the quality of geospatial or geo-coded data from

    the sites varied widely. The most problems we commonlyencountered included missing, inadequate, incorrect,inconsistent, and abbreviated geo-codes. We checkedand rechecked locations and family routines and thegeographic data sets on which these were mapped.When necessary, ethnographers returned to families andneighbourhoods to verify data points and familiesexplanations of their routines and experiences. We werefortunate that the structure of the project allowed for theGIS team members to query the data collectors themselvesand reconcile the most egregious data issues.

    It is worth returning briefly to the issue of privacy and therisks of identifying individuals. The ethnographic datafiles include information on the location of key anchorpoints such as home, workplaces, grocery stores, andpaediatricians offices. The GIS programmer worksdirectly with address, typically cross-street, information,removing all identifying information (i.e., any family ID,if it existed) from geospatial data files once the files aregeo-coded. As implied above, we use geo-coded data tovisualize family activity spaces over various time scales(typical daily, weekly, or monthly journeys) and geogra-phies (neighbourhood, town, county, study area), butwhen we do so the maps are devoid of any referencingdata such as streets and are reproduced only at smallscales, preventing identification of home addresses orother key anchor points of an individual or family. Accessto raw data for this project is limited, as data are housedin heavily restricted data directories within a Karlbridgefirewallprotected computer network.

    A strategy we have pursued for both training andgenerating interest among field ethnographers in collect-ing geospatial data is to visit their sites and present tothem material derived from their own field notes. Thishas allowed us to draw attention to the need for accurate,complete, and consistent data and provides an opportu-nity for the ethnographers to view the lives of theirfamilies from a different perspective; the process can alsotrigger discussions of the opportunities and constraintsfaced by the family. As field notes are often quiteunwieldy, some family ethnographers now generatespecific geospatial data files in spreadsheets. Indeed, incurrent work on low-income families we are usingencrypted Web forms to collect and organize data onlocations, routes, and boundaries and specificattributes about these location vis-a-vis our families. Thelocational data can be either direct or indirect geo-codes,depending on the specificity of the information requested(i.e., we may collect street addresses, cross-streets, ZIPcode fields, place names, or some other geographicdescriptor). Data from the forms are organized into adatabase file (including numerical and text string datafields) that can be easily converted to a format usable in aGIS system. A quick scan of the data by an ethnographercan reveal holes in the data.

    Geo-ethnography: Coupling Geographic Information Analysis Techniques with Ethnographic Methods in Urban Research

    cartographica (volume 40, issue 4) 87

  • It should be stressed that site visits to the three citiesbenefit the GIS team too. Our understanding of families issignificantly enhanced through one-on-one discussionswith field ethnographers and visits to the families homesand neighbourhoods. In each city, we drive and walkthrough study neighbourhoods, visit anchor institutionsand recruitment sites, and meet study families. Theseneighbourhood tours are invaluable, and we hope thatby better knowing and understanding our families studysites we can represent them more appropriately inmapped form.

    There have been some encouraging developments regard-ing the coupling of ethnography and GIS. First, a numberof research scientists and data coders (graduate studentsfrom a diverse set of social and health sciences) havealready taken formal GIS courses, or have plans to do so.That is, team members are beginning to understand betterhow geospatial data and attribute data (of various kinds)can be integrated and then queried within a GIS. Second,the research scientists and data coders (including thosewho have not used GIS) are increasingly sensitive to anarray of issues relating to the nature and quality of thegeospatial data they encounter in field notes. The datacoders have seen how other team members use the dataand thus increasingly recognize when the data may beincomplete or inaccurate. In these instances, a query to afamily or neighbourhood ethnographer at the specific sitecan help complete or correct information.

    Summary

    This article has focused on the coupling of geographicinformation system (GIS) technologies with ethnographicdata gathered in an ongoing, multi-site study of low-income families and their children, Welfare Reform,Children and Families: A Three City Study. Our goalshave been to think creatively about how GIS can be usedin welfare research, to stretch the technology, and to revisethe methodologies we currently use. Combining ethno-graphic data and GIS methods enables researchers to seespatial context as well as content, helping them to identifyimportant spatial dimensions of a problem. The couplingof ethnography and GIS techniques (i.e., geo-ethnogra-phy) has much to offer to the study of low-incomewomen and children in community contexts. We cansituate families in broader geographic contexts andvisualize the different strategies they employ in their useof neighbourhood resources, anchor institutions, andsocial networks. Moreover, we can compare and contrastthe use of neighbourhood resources and social networksamong families residing in the same neighbourhoods.By studying the use of resources and social networks,we hope to shed light on the complex and reciprocalrelationships between families and neighbourhoods. Theintegration of data on neighbourhoods and how families

    use spaces and places necessitates critical discussion ofhow neighbourhoods are operationalized in conventionalsocial science research. The ethnography reveals theactivity spaces and boundaries of a familys livedexperiences, which often extend well beyond the fixedspatial contexts of census tracts or neighbourhoodsused in more traditional quantitative approaches to thestudy of welfare research found in demography, sociology,and urban geography.

    In the Welfare Project, the emphasis is squarely on non-spatial areas of inquiry. However, as has been shown here,and as noted by Michael Goodchild and others (2000,154), even in non-spatial areas of inquiry an increasedawareness of the concepts and tools of spatial analysismay lead to novel insights. In closing, we strongly echothe comment made by Steve Herbert (2000, 564): Ifsociality and spatiality are intertwined, and if theexploration of this connection is a goal of geography,then more ethnography is necessary.

    Acknowledgements

    The authors wish to thank the entire ethnographic team fortheir contributions to this research endeavour. Specialthanks are due to the GIA core staff and graduate researchassistants at Penn State, who have worked with differentdata pieces, sat through various presentations, and madesuggestions that make our work better. Two colleagues areworth special mention. Alan Benjamin, the ethnographycoordinator on the Welfare, Children and Families Project,has provided valuable feedback and encouragement; DebraSkinner (UNC Chapel Hill) has collaborated in the integra-tion of disability ethnography with GIS, and many of thelessons we have learned are derived from that experience.Finally, we thank the organizers and participants of the GISand Critical Geographic Research conference held at HunterCollege, New York, on 25 February 2001, where the firstpublic presentation of our ideas was made. This article wasmade possible through support to the Population ResearchInstitute (NICHD Population Research Center Grant, R24 HD41025) and the Welfare, Children and Families Projectthrough grants from NICHD (R01 HD36093-02), the John D.and Catherine T. MacArthur Foundation, the W.K. KelloggFoundation, the Robert Wood Johnson Foundation, theJoyce Foundation, the Office of the Assistant Secretaryfor Planning and Evaluation, the Administration onDevelopmental Disabilities, the Administration for Childrenand Families, and the Social Security Administration.

    Author Information

    Stephen A. Matthews is Associate Professor of Sociology,Anthropology and Demography and Senior ResearchAssociate and Director, Geographic Information Analysis

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    88 cartographica (volume 40, issue 4)

  • Core Population Research Institute, The Pennsylvania StateUniversity.

    James E. Detwiler is GIS Analyst and Population ResearchInstitute and Faculty Instructor, Department ofGeography and World Campus Certificate Program inGIS, The Pennsylvania State University.

    Linda M. Burton is Principal Investigator, Welfare,Children and Families Project; Director, Center forHuman Development and Family Research in DiverseContexts, and Professor of Human Development andFamily Studies, Sociology and Demography, ThePennsylvania State University.

    Please address all correspondence regarding this article toStephen A. Matthews, Department of Sociology, ThePennsylvania State University, 211 Oswald Tower,University Park, PA 16802 USA. Tel.: (814) 863-9721.Fax: (814) 863-8342. E-mail: matthews@pop.psu.edu.

    Notes

    1. For an excellent review of the history of the welfare statein the United States, see Katz (2001).

    References

    Abler, R., J. Adams, and P. Gould. 1971. Spatial Organization:The Geographers View of the World. Englewood Cliffs, NJ:Prentice Hall.

    Aitken, S.C. 1998. Family Fantasies and Community Space.New Brunswick, NJ: Rutgers University Press.

    Booth, A., and A.C. Crouter, eds. 2001. Does It Take a Village?Community Effects on Children, Adolescents, and Families.Mahwah, NJ: Lawrence Erlbaum.

    Brooks-Gunn, J., G.J. Duncan, and J.L. Aber, eds. 1997a.Neighborhood Poverty, volume 1: Context and Consequencesfor Children. New York: Russell Sage.

    . 1997b. Neighborhood Poverty, volume 2: PolicyImplications in Studying Neighborhoods. New York: RussellSage.

    Burton, L.M., and J.E. Graham. 1998. Neighborhood Rhythmsand the Social Activities of Adolescent Mothers. In TemporalRhythms in Adolescence: Clocks, Calendars, and theCoordination of Daily Life, ed. R. Larson and A.C. Crouter.San Francisco. Jossey-Bass. 722.

    Burton, L.M., T.R. Hurt, C. Eline, and S.A. Matthews. 2001.The Yellow Brick Road: Neighborhoods, the Homeplace, andLife Course Development in Economically DisadvantagedFamilies. Paper read at the Second Biennial Conference ofthe Society for the Study of Human Development, 1921October, Ann Arbor, MI.

    Burton, L.M., R.L. Jarrett, L. Lein, S.A. Matthews, J. Quane,D. Skinner, C. Williams, W.J. Wilson, and T.R. Hurt. 2001.Structured Discovery: Ethnography, Welfare Reform, andthe Assessment of Neighborhoods, Families, and Children.Paper read at the Society for Research in Child Development,2023 April, Minneapolis, MN.

    Burton, L.M., and R.L. Jarrett. 2000. In the Mix, Yet on theMargins: The Place of Families in Urban Neighborhood andChild Development Research. Journal of Marriage and theFamily 62: 111435.

    Burton, L.M., S.A. Matthews, F. Avenilla, J. Quane, D. Graefe,and A. Benjamin. 2000. Ethnography, GeographicInformation Analysis, and Minority Family Health:Navigating Space, Place, and Health Events. Paper read atthe NIH Conference Towards Higher Levels of Analysis:Progress and Promise in Research on Social and CulturalDimensions of Health, 2728 June, Bethesda, MD.

    Coulton, C.J. 1997. Potential and Problems in DevelopingCommunity-Level Indicators of Childrens Well-Being.In Indicators of Childrens Well-Being, ed. R.M. Hauser,B.V. Brown, and W.R. Prosser. New York: Russell Sage.37291.

    Cromley, E.K. 1999. Mapping Spatial Data. In MappingSocial Networks, Spatial Data, and Hidden Populations, ed.J.J. Schensul, M.D. LeCompte, R.T. Trotter II, E.K. Cromley,and M. Singer. Walnut Creek, CA: Altamira Press/Sage.51124.

    Cromley, E.K., and S.L. McLafferty. 2002. GIS and PublicHealth. New York: Guilford.

    Curry, M.R. 1997. The Digital Individual and the PrivateRealm. Annals of the Association of American Geographers87: 68199.

    . 1998. Digital Places: Living with GeographicTechnologies. New York: Routledge.

    Furstenberg, F.F., and M.E. Hughes. 1997. The Influence ofNeighborhoods on Childrens Development: A TheoreticalPerspective and a Research Agenda. In Indicators ofChildrens Well-Being, ed. R.M. Hauser, B.V. Brown, andW.R. Prosser. New York: Russell Sage. 34671.

    Gahan, C., and M. Hannibal. 1998. Doing QualitativeResearch with QRS NUD.IST 4. London: Sage.

    Gephart, M.A. 1997. Neighborhoods and Communitiesas Contexts for Development. In Neighborhood Poverty,volume 1: Context and Consequences for Children, ed.J. Brooks-Gunn, G. Duncan, and J.L. Aber. New York: RussellSage. 141.

    Gilbert, M.R. 1997. Identity, Space, and Politics: A Critique ofthe Poverty Debates. In Thresholds in Feminist Geography:Difference, Methodology, Representation, ed. J.P. Jones III,H.J. Nast, and S.M. Roberts. Lanham, MD: Rowman &Littlefield. 2946.

    . 1998. Race, Space, and Power: The Survival Strategiesof Working Poor Women. Annals of the Association ofAmerican Geographers 88: 595621.

    Golledge, R.G., and R.J. Stimson. 1997. Spatial Behavior:A Geographic Perspective. New York: Guilford.

    Goodchild, M.F., L. Anselin, R.P. Applebaum, andB. Herr Harthorn. 2000. Toward Spatially IntegratedSocial Science. International Regional Science Review 23/2: 13959.

    Hagerstrand, T. 1970. What About People in RegionalScience? Papers of the Regional Science Association 24:721.

    Herbert, S.K. 2000. For Ethnography. Progress in HumanGeography 24: 55068.

    Geo-ethnography: Coupling Geographic Information Analysis Techniques with Ethnographic Methods in Urban Research

    cartographica (volume 40, issue 4) 89

  • Katz, M.B. 2001. The Price of Citizenship: Redefining theAmerican Welfare State. New York: Metropolitan.

    Kwan, M.-P. 1999. Gender and Individual Access to UrbanOpportunities: A Study Using Space-Time Measures.Professional Geographer 51: 21027.

    . 2000. Interactive Geovisualization of Activity TravelPatterns Using 3-D GIS. Transportation Research Part C 8:185203.

    . 2002a. Time, Information Technologies, and theGeographies of Everyday Life. Urban Geography 23: 47182.

    . 2002b. Feminist Visualization: Re-Envisioning GIS as aMethod in Feminist Geographic Research. Annals of theAssociation of American Geographers 92: 64561.

    Laws, G. 1997. Womens Life Courses, Spatial Mobility, andState Policies. In Thresholds in Feminist Geography:Difference, Methodology, Representation, ed. J.P. Jones III,H.J. Nast, and S.M. Roberts. Lanham, MD: Rowman &Littlefield. 4764.

    Leventhal, T., and J. Brooks-Gunn. 2000. The NeighborhoodsThey Live In: The Effects of Neighborhood Residence onChild and Adolescent Outcomes. Psychological Bulletin 126:30937.

    Martinez, M.L. 2000. Neighborhood Context and theDevelopment of African American Children. New York:Garland.

    Matthews, S.A., L.M. Burton, and T.R. Hurt. 2002. CombiningGIS and Ethnography to Explore Social Networks andNeighborhood Resource Use among Low-Income Families.Paper read at the annual meeting of the PopulationAssociation of America, 911 May, Atlanta, GA.

    Matthews, S.A. 2005. What Do We Know about Families andNeighborhoods from Geoethnography. Unpublished manu-script, available from lead author.

    McLafferty, S.L. 2002. Mapping Womens Worlds:Knowledge, Power and the Bounds of GIS. Gender, Placeand Culture 9: 26369.

    Monmonier, M. 2002. Spying with Maps: SurveillanceTechnologies and the Future of Privacy. Chicago: Universityof Chicago Press.

    Pickles, J. 1995. Ground Truth: The Social Implications ofGeographic Information Systems. New York: Guilford.

    Rose, G. 1993. Feminism and Geography: The Limits ofGeographical Knowledge. Minneapolis: University ofMinnesota Press.

    Rowe, S., and J. Wolch. 1990. Social Networks in Timeand Space: Homeless Women in Skid Row, Los Angeles.Annals of the Association of American Geographers 80/2:184204.

    Sampson, R.J., J.D. Moronoff, and F. Earls. 1999.Beyond Social Capital: Spatial Dynamics of CollectiveEfficacy for Children. American Sociological Review 64:63360.

    Sastry, N., A. Pebley, and M. Zonta. 2002.Neighborhood definitions and the spatial dimensionsof daily lives in Los Angeles. Paper read at the annualmeeting of the Population Association of America, 911 May,Atlanta, GA.

    Skinner, D., S.A. Matthews, and L.M. Burton. 2005.Combining Ethnography and GIS Technology to ExamineConstructions of Developmental Opportunities in Contextsof Poverty and Disability. In Discovering SuccessfulPathways in Childrens Development: Mixed Methods inthe Study of Childhood and Family Life, ed. T.S. Weisner.Chicago: University of Chicago Press/MacArthur Foundation.22339.

    Winston, P., R.J. Angel, L.M. Burton, P.L. Chase-Lansdale,A.J. Cherlin, R.A. Moffitt, and W.J. Wilson. 1999. Welfare,Children and Families Three City Study: Overview and DesignReport. Baltimore, MD: Johns Hopkins University, Welfare,Children and Families Project.

    Welfare, Children and Families. 2002. Welfare, Children andFamilies home page. Available at http://www.jhu.edu/welfare.

    Stephen A. Matthews, James E. Detwiler, and Linda M. Burton

    90 cartographica (volume 40, issue 4)

    JOURNAL ARTICLE 05-09

Recommended

View more >