Taking up space in educational research by Victor Sensenig

Educational researchers are taking an increasing interest in social space, how people use and attach meanings to the “stuff” of natural space. In the past three years, the American Journal of Education has published ten articles that directly consider the effects of spatial factors on educational processes, six of them in a special issue called “Mapping Educational Opportunity.” The concept of “spatial capital” has been coined to describe unequal control over space and the ways that space contributes to social reproduction, as when families’ location determines their children’s school provision (Barthon and Monfroy 2010). What led to the spatial turn in educational theory and policy, and where is this turn taking us?

There are two reasons for this growing interest in space. First, many of the most pressing educational issues of the last fifty years have had a basic spatial dimension. The big story in school segregation after the Brown v. Board of Education decision was the shift in segregation within neighboring schools to a pattern of city-suburban segregation. The school choice debate also pivots on spatial questions, including whether students should be freed from their historical assignment to a local public school and who would benefit from an unmoored, market-based configuration.

At an even more fundamental level, the physical space people occupy can extend or deny their life-chances, including their opportunity to learn. Exposure to neurotoxins like lead is highly dependent on where people live, contributing to differences in cognitive development (Sanders, Liu, Buchner, and Tchounwou 2009). Beneficial environmental factors, such as print exposure, are also unevenly dispersed over space (Neuman and Celano 2001). Spatial analysis holds great promise for understanding the “geography of opportunity” (Briggs 2005).

A second reason for educational interest in geography has been the tremendous recent advances in the tools used for spatial analysis, particularly geographical information systems (GIS), and increases in the quantity of available information. GIS involves the integration of spatial coordinate data and non-spatial attribute data and enables some powerful analysis techniques. Set theory, including operations of intersection and union, can be extended by map algebra to calculate new attribute values for spatial areas. For example, a researcher can create a map that shows where schools are located relative to different neighborhoods in a city and then transform the data into new information in order to answer particular queries. For example, what are the per pupil expenditures of schools in neighborhoods that have a minority population of over 50 percent? Answers to questions like this one can help us to understand the contribution of segregated living to educational inequality.

However, in order to represent geographic phenomena in a data structure, geographic models of reality require certain assumptions. These assumptions must be acknowledged when spatial factors are used in educational research. One use of space, especially in the economics of education literature, has been its exploitation as an instrumental variable. In one study, Black (1999) investigated the question of the effects of school quality on house prices. Regression results confront the problem of omitted neighborhood characteristics, confounding the conclusion that better schools lead to higher spending on homes. Black devised an ingenious solution, limiting her sample to houses near an attendance boundary. This technique minimized neighborhood differences while ensuring that differences in elementary school quality remain a significant factor in a home-buying decision. She could then attribute differences in house prices across these boundaries to differences in school quality.

Distance to school is also a popular instrumental variable. Cullen, Jacob, and Levitt (2004) attempted to determine the consequences for graduation rates of students who opted out of their assigned school after the Chicago Public Schools instituted open enrollment. Students who opted out would be expected to differ from those who didn’t in important unobserved dimensions of motivation and parental involvement, but this study uses the distance to school counter the upward bias of the estimate of increased graduation rates.

“Ultimately, space matters in the study of education primarily because it affects children’s chances to interact with caring, knowledgeable people…”

These uses of space also do not try to understand how it works but instead leverage it to answer other questions. In both of these cases, the assumption that allows space to be used to clear certain statistical hurdles is the utter abstraction of distance, its removal from a social context for use as a “quasi-Newtonian” independent variable (Soja 1989, 149). Distance can be used as an instrumental variable because it is assumed to exert a predictable friction. It can only be crossed with time, money, and effort, so it reliably diminishes interactions across space as it increases. This use of distance is challenged by evidence that the effects of distance are contingent on place and that social space defies reductive quantification.

Another major way of considering space is to outline zones and then assign particular characteristics to people in those zones. Sohoni and Saporito (2009) compared the racial composition of public schools and school catchment areas, addressing the critical school choice policy question of how enhanced individual school choice affects segregation. GIS enabled them to assign block-level Census data to school attendance areas in order to determine their racial composition. They then located schools and their enrollment data within these boundaries. The local schools in their sample are indeed more segregated than their catchment areas, suggesting that non-neighborhood private, magnet, and charter schools exacerbate segregation. When they are given choices, these families choose more homogenous schools than the areas where they live.

In another study, Koontz, Jue, and Bishop (2009) analyzed the demographics of geographic areas within a one-mile radius of 134 permanently closed public libraries. They accomplished this by overlaying geocoded library addresses on Census 2000 data. They find that library closures are associated with lower rates of home ownership, lower income, and lower educational attainment than national averages. The implications of these two studies’ findings are that when we draw lines around the geographical areas were people live and learn, we can detect patterns of the inequitable distribution of goods or undesirable social sorting. The problem for these studies derives from how these lines are drawn.

The critical assumption made in these studies is that Census block data describe uniformly distributed individuals. In effect, individuals are assigned the average income of their Census block and are assumed to be evenly dispersed. In the Koontz et al. (2009) study, the circle that the researchers draw around each closed library cuts through Census blocks, and individuals are assigned to the included portion according to its proportion of the whole block. This is not always a reasonable assumption, and this approach ignores the fact that these data are created by the assignment of fields with continuous variation to arbitrarily bounded zones. This approach assumes the internal homogeneity of the zones, which are derived from administrative fiat rather than from more significant boundaries in the socioeconomic geography of an area. Census data is widely used because it is readily available, but zone-based data always confront this ecological fallacy (Martin 1996).

A third approach recasts space as place, building on the insight that people can perceive the same space in different ways. According to Bell (2009), “Place refers to the social, economic, and political meanings people assign to particular spatial locations” (495). Bell’s study of 48 families in Detroit attempted to understand how parents’ geographical preferences affected their decisions about where their children would attend school. She distinguished between choice sets—the schools that parents considered—and geographic sets, or the schools within a two-mile radius of their residence. Choice sets were determined by the meaning that parents assigned to particular neighborhoods and schools in tagging them as “safe,” or as a “good fit.” Bell concluded that the parents’ perceptions of particular places exerted a powerful influence over their decisions and that geographic constraints operated in unexpected ways. The concept of convenience can be easily conflated with simple distance but it is more complex than that, involving factors like the availability of transportation and work schedule negotiations.

Bell’s place-based notion of geography infuses a qualitative dimension into conventional quantitative geospatial analysis, but the study still posits a “real” geography as a backdrop for its participants’ subjective representations of the world. Its findings derive from the intriguing differences between these versions of the world. GIS visualization is not how people see and experience the world: Bell’s study points to ways in which geospatial data could be represented in a more embodied way. Along these lines, one method that could prove useful to educational researchers incorporating geospatial analyses is Kwan’s (2007) “daily space-time paths,” which trace individuals’ movements over space and time using GIS but also incorporate qualitative data produced by participants, including handwritten notes and maps, digital photos, and audio clips. Kwan then uses a method of “3D GIS videography” to describe, for example, the experience of Muslim women in the U.S. after 9/11. She creates a film that color-codes the spaces and buildings according to participants’ perceptions of danger in their daily travel. In this approach, space does not act on people in a uniform way but is interpreted by them. This way of thinking about space relies on the insight that, as Urie Bronfenbrenner has argued, “the environment of greatest relevance for the scientific understanding of behavior and development is reality not as it exists in the so-called objective world but as it appears in the mind of the person” (23).

This overview suggests that educational researchers must adapt the tools and techniques of geospatial analysis to the purposes of a science of learning. Geographic models using GIS do not provide an account of why space matters in education. They are agnostic regarding theories of human learning and development and can be susceptible to deterministic or behavioristic interpretations as well as to bird’s eye views of the world. Certainly, some characteristics of the spaces people occupy act on them in relatively straightforward ways, as in the inexorable, tragic consequences of a child ingesting lead dust over several years. However, people live in complex human and physical environments, which exert influence through both physiological and psychological factors, and we requires theories that can cope with this complexity.

As one promising candidate, ecological systems theory, points out, children are affected by conditions and decisions that are spatially remote, such as parents’ workplaces and state legislatures. In addition, individuals are embedded in environments but also have transactional relationships with their environments. Ultimately, space matters in the study of education primarily because it affects children’s chances to interact with caring, knowledgeable people in meaningful and increasingly complex activities and to avoid the people who could do them harm.

References

Barthon, Catherine, and Brigette Monfroy. 2010. “Sociospatial Schooling Practices: A Spatial Capital Approach.” Educational Research and Evaluation 16 (2): 177-196.

Bell, Courtney. 2009. “Geography in Parental Choice.” American Journal of Education 115 (4): 493-521.

Black, Sandra E. 1999. “Do Better Schools Matter? Parental Valuation of Elementary Education.” The Quarterly Journal of Economics 114 (2): 577-99.

Briggs, Xavier de Souza. 2005. The Geography of Opportunity: Race and Housing Choice in Metropolitan America. Washington, DC: Brookings Institution Press.

Bronfenbrenner, Urie. 1979. The Ecology of Human Development: Experiments by Nature and Design. Cambridge, MA: Harvard University Press.

Cullen, Julie B., Brian A. Jacob, and Steven D. Levitt. 2005. “The Impact of School Choice on Student Outcomes: An Analysis of the Chicago Public Schools.” Journal of Public Economics 89 (5): 729-760.

Kwan, Mei-Po. 2007. “Hybrid GIS and Cultural Economic Geography.” In Politics and Practice in Economic Geography, edited by A. Tickell, E. Sheppard, J. Peck, and T. Barnes. London: Sage.

Martin, David. 1996. Geographic Information Systems: Socioeconomic Applications. London: Routledge.

Neuman, Susan, and Donna Celano. 2001. “Access to Print in Low-Income and Middle-Income Communities: An Ecological Study of Four Neighborhoods.” Reading Research Quarterly 36 (1): 8-26.

Sanders, Talia, Yiming Liu, Virginia Buchner, Paul B. Tchounwou. 2009. “Neurotoxic Effects and Biomarkers of Lead Exposure: A Review.” Reviews on Environmental Health 24 (1): 15–45.

Sohoni, Deenesh, and Salvatore Saporito. 2009. “Mapping School Segregation: Using GIS to Explore Racial Segregation Between Schools and Their Corresponding Attendance Areas.” American Journal of Education 115 (4): 569-600.

Soja, Edward. 1989. Postmodern Geographies: The Reassertion of Space in Critical Social Theory. New York: Verso.