1 Using Exploratory Spatial Data Analysis Techniques to Better Understand Housing Discrimination1 This paper explores the potential for mapping with Geographic Information System (GIS) technology and Exploratory Spatial Data Analysis techniques using several different software packages to contribute to an understanding of housing discrimination in the City of Philadelphia. The primary research question is whether spatial statistical analy-sis offers insight beyond that provided by visual analysis of point patterns and area data. Various K-functions are used to test for significant clustering and spatial dependence be-tween events while spatial autoregression and spatial lag programs are used to test for and model spatial autocorrelation. This varied approach leads to important conclusions concerning methodology as well as results relating to the spatial relationships among the location of different types of housing discrimination and to neighborhood characteristics including race and income. Throughout United States history, realtors, lenders, rental agents, residential managers, landlords, home sellers, newspaper publishers and governments at all levels have actively practiced discrimination that has limited access to housing for significant segments of the population (Yinger, 1995). Though numerous judicial and legislative policies were es-tablished in the 100 years before it, the federal Fair Housing Act (FHA) of 1968 provided the first comprehensive foundation for the identification and elimination of housing dis-crimination by explicitly outlawing discrimination on the basis of race, color, religion, or country of origin in the sale and rental of housing. The 1988 Fair Housing Act Amend-ments (FaHAA) built on this foundation by significantly increasing the minimum dam-ages for first time offenders, improving the administrative process through which the De-partment of Housing and Urban Development (HUD) handles complaints, and adding familial status and handicap to the list of protected classes. In spite of these legislative advances and the increased funding for fair housing activity that they have provided to states and municipalities, current research indicates overwhelmingly that discrimination remains entrenched (Yinger, 1995). As outright dis-crimination has been banned from public and private transactions, the techniques used to discriminate against tenants and homeowners have proliferated. More sophisticated re-search methods are needed to identify discriminatory practices as they become subtler. Testing and audit methodology, in which the treatment and housing outcomes offered to paired “testers” who have been matched on all but a single variable (such as race) are compared, have been developed into sound and effective investigative tools. But like other research methods, these do not reflect the inherently spatial nature of housing. Rather than pointing to the location of acts of discrimination and the relationship among the location of different incidents, traditional research methods seek to explain the phe-nomenon solely through individual characteristics of the victim and perpetrator. 1 This paper is based on data collected for a previous study by the author and David Eldridge, MSW that was made possible by the Housing Association of Delaware Valley and the William Penn Foundation.2 Much of social science research neglects spatial relationships, but this is particu-larly limiting in housing research because of housing’s unique emphasis on location. Where property is located has always been a primary issue in real estate. Redlining, the practice of not lending to particular areas based on location and racial composition, is based on location and literally involved drawing red lines on maps during its infancy (Jackson, 1993; Massey and Denton, 1993). The primary goal of this paper, then, is to demonstrate how the combination of Geo-graphic Information System (GIS) technology and exploratory spatial data analysis (ESDA) techniques offers a new approach to fair housing research by providing the tools to identify and understand spatial relationships. GIS uses computer software to integrate, analyze, and display data that are identified with some level of geography, be it specific addresses or areas such as census tracts. GIS is used here to map the location of incidents of housing discrimination in relation to neighborhood characteristics. ESDA techniques have a wide range of applications and are appropriate for exploring point patterns, con-tinuous spatial data, as well as area data. The techniques used here test for significant clustering among the locations of events and independence between different types of events as well as spatial autocorrelation. On their own, maps generated by GIS provide clear evidence of the clustering of incidents of housing discrimination and their associa-tion with neighborhood characteristics. The question to be explored in this paper, then, is whether ESDA contributes to an understanding of housing discrimination beyond that provided by simple—but powerful—visual analysis. BACKGROUND Laws against housing discrimination cover a wide variety of types of discrimination and vary at the state and local level according to the classes of people they are designed to protect. Fair housing issues refer to the specific discriminatory activity and answer the question, “How did discrimination happen?” Bases, on the other hand, refer to specific protected classes and answer the question, “Why did discrimination happen?” The opportunities to limit where protected classes of people choose to live are quite nu-merous and the tactics used range from blatant to extremely subtle. Refusing to show, sell or rent a property are common issues. Offering different terms and conditions (such as higher security deposit because of presence of children), different conditions for offer-ing services (such as repairs), refusing to make reasonable accommodations for people with disabilities, and evicting a tenant are some of the ways in which renters are dis-criminated against. Steering, the practice of showing certain potential renters or buyers some properties and not others, and blockbusting, or scaring current homeowners into selling with threats that changes on the block are imminent, are two of the ways real es-tate agents have discriminated. Retaliation against someone who has made a fair housing