THE WORKSHOP: GIS/GIA AND HUMAN CAPITAL RESEARCH

TO IDENTIFY THE SPECIFIC CONTRIBUTIONS THAT GIS/GIA CAN MAKE TO human capital research initiatives, three case studies will be used to highlight the potential benefits. These cases develop different subjects of relevance to the human capital research initiative (crime, social networks, and neighborhoods and demographic change), each characterized by a reliance on a different type of spatially referenced data (respectively, events; networks or linkages; and areas or polygons). As such, they facilitate a wide-ranging and thorough exploration of the limitations and strengths of the geographic perspective outlined above. The case studies reveal interesting parallels in the experiences of each working group, but also point to some significant differences. Collectively, they attest to the vitality of the geographic perspective on human capital topics and illustrate some of the potential applications of GIS/GIA for advancing research in this area. The working groups also uncovered some areas in which further intellectual and technological developments would enhance the ability of non-geographers to reduce the start-up costs of incorporating space into their research agendas. We return to these as recommendations later in the report.

After opening overviews and demonstrations on GIS/GIA, group leaders introduced their data at plenary sessions at the conference. While participants were assigned to one of the three groups (based on the statement of personal research interests each participant forwarded in advance of the meeting) for the first focus-group sessions, they were encouraged to change groups at midday. While many remained with their initial group assignments, several took advantage of this opportunity to participate in more than one group. The commentary on these cases is based on the experiences of focus groups organized around each case. This discussion will include a brief description of the substance and the data for each case, and the nature and direction of focus group questioning. It should be emphasized that these cases do not intend to exhaust the wide range of subjects that might be covered by scholars developing a spatial perspective on human capital research; rather, they are intended to be suggestive of the applicability of spatial perspectives in advancing understanding across a range of substantive questions and data types. In this respect, the groups performed well. The names of the group leaders who assembled the data and led the focus group sessions appear in parentheses.

a. Crime and Human Capital
(Professors Carol W. Kohfeld and John Sprague)

Violent crime is a distressing reality for many in contemporary America. A recent Department of Justice report forecast that 83 percent of Americans would become a victim of a violent crime at some time in their lives (The Atlantic Monthly, July 1995). This probability is spatially variant, however, for violent crimes exhibit a geographic distribution that is strikingly clustered. Inquiry into this geography of violent crime is obviously and significantly related to the research agenda examining the structures of disadvantage in American society. Insofar as criminal activity is concentrated in communities already afflicted by poverty and deprivation, understanding the geography of crime is central to this policy responses designed to improve the human capital stock of these disadvantaged communities.

For a number of years, Professors Kohfeld and Sprague have been working closely with the St. Louis police department on a project involving homicides and other serious crimes. In the course of this collaboration, they have been given access to police crime reports roughly covering the past thirty years. They have geocoded these criminal events, and have generated coverages of the city of St. Louis illustrating the spatial distribution of a variety of serious crimes and, in particular, homicides (incorporating information on victims, suspects, and the location of the criminal act). This aspect of the project illustrates an important issue that attends virtually every geographic analysis of sensitive data, namely the question of privacy and confidentiality. It is obvious that the address is a highly revealing piece of information and that the opportunities for linking and archiving data using addresses represents in potentiality, if not in practice, a threat to the privacy of individuals. In preparing this data for the conference, care was taken to destroy individual address information embedded in the electronic versions of the crime-data files. While many privacy issues can be dealt with on a case-by-case basis, there clearly is a need to address these concerns in the development of more general principles governing the storage and dissemination of spatially referenced event count and public opinion survey data (the census has developed guidelines governing the disaggregation of their data that might serve as a guideline in this respect).

The data that resulted from these procedures afforded an unprecedented opportunity to explore the spatial structures and correlates of violent crime. In addition, the St. Louis project leaders collected census and electoral data for the city since 1970. In this respect, they have taken advantage of the fact that the census geography of St. Louis has not significantly changed over the 1970-1990 time period. Thus, this group had access to a uniquely rich store of census data and information on criminal activity for essentially a constant set of geographic units. Participants had an opportunity to query the data in both GIS (ArcView) and statistical (STATA) formats.

Overlays of criminal events across time suggested considerable spatial stability in the areas where these incidents were concentrated. Adding layers representing selected demographic features of census tracts provoked some discussion of the likely causal factors associated with the incidence of homicides. One particularly striking visual correlation could be seen between tracts in which single, female-headed families were concentrated and the incidence of homicides, prompting some to hypothesize the intensity of familial supervision (or its weakness) might be associated with the propensity of individuals to commit this crime. Cartographic displays of homicides in St. Louis reveal a spatial clustering of these events, and discussion in this group focused (more than others) on the issue of modeling violent crime in the presence of spatial dependence among the units of observation. Additionally, the group discussed the challenges of incorporating a temporal dimension within analyses of the spatial structure of criminal events.

Based on the discussion and exploration of the data, the group offered the following summary observations/recommendations:

  • Collaboration between geographers and other social scientists leading to the publication of geographic analyses in non-geographic journals is the key to the penetration of GIS/GIA methods in the other social sciences. Since the geographic distribution of departments of geography is spatially sparse, a significant promotional role is available for those who fund GIS/GIA research or social scientific research, or both, in promoting productive collaborative work.

  • Even public agencies for whom GIS/GIA technologies are especially appropriate may ignore or be unaware of the analytic possibilities of such applications. A review of interagency use and support of GIS/GIA is recommended.

  • A central analytic issue was the difficulty of the marriage between areal and point data in databases containing each type of information. The direction of resolution of this issue probably lies in borrowing from the work of physical geography and the environmental sciences, where work with such "mixed" databases is more advanced.

  • Overwhelming agreement was expressed that a better integration between display technologies and analytic technologies would provide a cornerstone for progress in making GIS/GIA technologies available and useful to the social scientific community.

  • All agreed that the marriage of display and analytics for space and time jointly remains a difficult problem.

    b. Neighborhoods, Social Networks, and Human Capital
    (Professor Robert Huckfeldt)

    In collaboration with Professor John Sprague, Professor Huckfeldt assembled a three-wave survey of residents of sixteen neighborhoods in South Bend, Indiana, at different points in the campaign year of 1984. An important and innovative feature of this rich survey was the inclusion of a "snowball" sample of discussion partners named by the primary survey respondents. The surveys covered a wide range of issues concerning the attitudes, associational and group ties, and political behavior of respondents. The spatial- boundedness of social relationships is an important factor contributing to the strength and character of residential neigborhoods. Where neighborhoods define meaningful and positive communities of interaction for their residents, such communities can be seen to enjoy relatively high levels of social capital (and vice versa). The extent to which strong neighborhoods, as repositories of social interactions and social capital, encourage positive human capital decisions among their residents constitutes an important aspect of human capital research.

    In preparation for the Boulder meeting, the addresses of primary respondents and their discussion partners (when the partner named was someone other than a spouse) were geocoded into ArcView coverages. In addition, census data for South Bend was obtained from the 1990 census and appended to the survey data. Participants in this group could query this data using GIS (ArcView) and statistical (SYSTAT) formats. This combination of data enabled the group to explore the geography of political discussions, social networks, and human/social capital endowments of South Bend neighborhoods. As Robert Putnam's recent work has exemplified, such informal social processes are reflective of the social capital stock of neighborhoods and are significantly related to a variety of measures of civility, effective governance, and communal vitality.

    Discussion in this group focused on two general issues. First, participants agreed that there was great potential for GIS/GIA technology to contribute to the difficult task of defining neighborhood boundaries and to explore the various ways in which individuals might construct different conceptions of neighborhoods for different purposes. It was agreed that the spatial structure of social networks might be one factor that should be considered in defining neighborhoods, but that other factors might also play a role. A second area of discussion centered on the usability of GIS/GIA by non-geographers in the social sciences. There was some concern that the start-up costs involved in mastering a different set of software skills, and in preparing spatially referenced social science data sets, might discourage some scholars from taking advantage of this perspective.

    Following Professor Huckfeldt's presentation, participants inquired about the extent to which the sixteen different primary-respondent neighborhoods were characterized by networks with different spatial structures (represented by the distance separating main respondent and discussion partner). Displays of the network dyads (represented by lines on a map of South Bend joining respondent and discussion partner) were generated and compared for a number of neighborhoods, and descriptive statistics for the mean distance were compared across all study neighborhoods. Some interesting differences resulted and occasioned some discussion. The spatial structure of the networks for well-educated and less-well educated respondents were also compared, both visually and statistically.

    Based on these proceedings, the group offered the following observations and recommendations:

  • GIS has enormous potential for assisting in the visualization of many forms of data, and may well be highly appropriate for assessing the spatial boundedness of a variety of social phenomena we typically associate with neighborhoods. Different social areas may well be appropriately described as neighborhoods reflecting different purposes (e.g., shopping, socializing, perceptions of belonging and identity, etc.).

  • GIS is probably more limited as a vehicle for capturing the complexity of many of these social processes. The statistical techniques of GIA may be more helpful in this respect. GIS, however, may have an important role to play in preparing social science data sets for an analysis that incorporates a significant spatial dimension in more sophisticated modeling efforts. Participants encourage GIS software developers to enhance the analytic and statistical capabilities of existing packages.

  • If the full potential of GIS/GIA as a general tool for social scientists is to be realized, some provision for introductory and advanced training ought to be made, perhaps in the form of a central repository of expertise in social scientific applications of these tools/techniques.

    c. Migration, Demographic Change, and Human Capital
    (Professors Peter Rogerson and Richard Morrill)

    One of the most significant life decisions that individuals make is the decision to change residences. The motivations for residential mobility are many, but one important determinant of this decision is the structure of economic opportunities (local and extra-local). As such, residential change is one aspect of an individual's human capital decision making that relates to the matching of employment and skills in an economy. An aggregate consequence of these individual decisions to relocate is demographic change, a process which dramatically affects the distribution of human capital endowments across space. In preparation for the Boulder meeting, census data on neighborhood change within the city of Buffalo, New York, for the 1980-1990 period had been assembled in ArcView format. The decade of the 1980s was a period of major demographic change in this city, reflecting the out-migration of thousands of families - Buffalo now is smaller in population terms than it was in 1900. The data enabled participants in this group to explore the consequences of out-migration on the demography (human and social capital endowments) of Buffalo neighborhoods.

    After an initial presentation on the macrolevel features of population mobility by Professor Morrill, discussion in this group focused on the central theoretical and practical questions of how space matters in understanding human capital decisions. Migration patterns involve a complex mix of local and regional (and even international) dimensions. One of the ways in which GIS/GIA was found to be relevant to these issues was in addressing the issues of the multiple (spatial) scales that affect the human capital decisions of individuals. In addition, participants inquired into the measures and meaning of neighborhood stability/instability, and agreed that GIS/GIA could make a useful contribution to the mapping of "opportunity areas" for individuals and to the visualization of patterns of neighborhood stability and change, in particular as regards such externalities as pollution and congestion. Some discussion also centered on the issue of the role of GIS/GIA in understanding the processes of neighborhood change associated with the age profile of residents.

    Based on the activities and discussion, the group offered the following summary observations and recommendations:

  • GIS/GIA has considerable potential to contribute to human capital research in a variety of ways, such as visualizing the spatial distribution of social, political, economic, and other forms of opportunity and distress.

  • GIS/GIA can help identify natural or artificial barriers to access and identify choke points in the provision of public services.

  • The ability to shift geographic scales of analysis using GIS is helpful in exploring for the most appropriate level of analysis at which to pitch the search for understanding.

  • Further effort should be made to integrate the analytical and display capabilities of GISs. Participants agreed that this limitation in currently available software constitutes a significant impediment to the widespread adoption of GISs in the social and behavioral sciences.


    Forward Advancing a Geographic Perspective on Human Capital Research

    Backward Geographic Perspectives and Human Capital Research

    Table of Contents