Lawrence A. Brown
Department of Geography, The Ohio State University, Columbus

Position Statement
Curriculum Vitae
Address

Position Statement

THE GIS/SA INTERFACE FOR SUBSTANTIVE RESEARCH(ERS): A CRITICAL NEED


A central question of this conference is "how successful GIS has been at making spatial analysis widely available to physical and social scientists".  To put the question somewhat differently, to what degree is there an intersection of GIS and Spatial Analysis which maximizes the power of this combination; facilitates its use by scientists who are focussed on substantive, empirical-based research rather than methodology per se; and propagates such use by example, instruction, and the like?

In my estimation, and speaking as the type of person described in the question, an end-user carrying out empirical research, the answer is "not enough" and the ongoing scenario is not encouraging.  Put another way, I believe there is a tremendous need for classroom training, workshops, published examples that can serve as guideposts, and software development -- all of which would facilitate the use of analytic methods that exploit the power of the GIS/SA interface among a broad range of end-users with varying levels of, or inclination towards, methodological expertise.

Addressing this need can make a significant (monumental)  difference in the standing of Geography among other sciences.  It also can be a significant catalyst towards shifting current research trends/tendencies that too often neglect empirical analysis, particularly of a statistical nature.

Discussion around this broad point can be divided into 6 topics.  (1) Parallels with the Quantitative Revolution; (2) Types of GIS Efforts; (3) Gains from the GIS Revolution; (4) GIS/Spatial Analysis in Geography and Available Analytical Tools; (5) Who uses GIS/SA in Scholarly Research that is Substantively Motivated; (6) What is the problem, how can it be remedied?  In addressing this topic, I do so as an end-user Geographer concerned with advancing the understanding of academic issues central to our respective disciplines; solving problems by applying GIS/SA tools is an important task, but not among those I identify as central in the context of this conference.

(1) Parallels with the Quantitative Revolution

A useful parallel is the "quantitative revolution" of the 1960s and 1970s. Initial research efforts were focussed on learning to use techniques developed by others, developing and/or adapting techniques particularly relevant to spatial analysis, writing computer code to ease application, and eventually, adopting generally available statistical software such as the BIOMED, SPSS, and SAS packages.  While quantitative analysis struggled for acceptance initially, by 1980-85 it was both widely used and seen as essential to academic training at both the graduate and undergraduate levels.  Perhaps the single-most important catalyst in this (together with empirical research that demonstrated the value of quantitative analysis) was SPSS and SAS that made statistical analysis readily accessible and included methodological instruction together with easy-to-use software.

In a similar fashion, GIS has been propelled forward by software such as ArcView, which is considerably more user-friendly than ArcInfo (at the sacrifice, at least momentarily, of analytical power).  The GIS/SA interface has been enhanced by packages such as SPACESTAT and S-PLUS, but user-friendliness has been an issue. There are indications that this shortcoming is now being addressed and user-friendliness is on the upswing.  However, attention to the issue appears to be a recent occurrence.

The intersection of academic courses is another issue that arises by considering quantitative revolution parallels.  That involved methods courses, but also use of those methods in substantive courses and published research that provides a "demonstration effects".  In academic settings today, GIS courses are generally separate from Spatial Analysis (usually statistics) courses;  reference to, or use of, these in substantive courses has fallen measurably; and this is particularly true of GIS and SA together, the GIS/SA interface.  In my opinion, an important force here is the limited use of GIS/SA in substantive academic research and/or confinement of its use to a small set of substantive research problems.

(2) Types of GIS Efforts

Elsewhere (September 1996 President's Column, AAG Newsletter, "The G in GIS -- Getting It Right") I postulated that GIS, from the perspective of Geography, involves three types of endeavors --

  Routine-Descriptive GIS, using GIS software to make maps,   diagrams, and the like;

  Analytical-System Design GIS, joining GIS with statistics,   cartography, information retrieval, and similar tools to   answer substantive questions of a scholarly or applied nature;

  Technical-System Development GIS, advancing GIS software,  analytic systems, etc.

My strong feeling is that the competitive advantage of Geography lies in the second of these, the GIS/SA interface and its application to substantive research.

(3) Gains from the GIS Revolution

There have been many gains from the GIS Revolution -- NSF's initiative towards a National Center for Geographic Information Analysis (NCGIA); Ohio State's Center for Mapping; GIS as a central ingredient in government and business efforts involved with environmental management, urban planning, facility location, marketing, transportation logistics, and the like; the numerous GIS software companies and business ventures (e.g., ESRI, Business Geographics Conference, "Business Geographics" and "GIS World" publications); a geographer at the helm of the US's Census 2000.

But in my mind, two specific changes in recent years highlight the impact of the GIS Revolution.  First, the visibility of Geography, as a discipline, has been tremendously enhanced. Second, maps and the portrayal of information in map form has taken on new meaning for the population at large, not to mention scientific endeavors outside of Geography.

This raises a vital question.  What should Geography do, now and in the future, to get the G in GIS right so as to position ourselves for the next epoch -- to leverage and build our present advantage such that GIS gains are not simply a wave that passes?

(4) GIS/Spatial Analysis in Geography and Available Analytical Tools

My observation is that the use of GIS/SA is highly bifurcated -- (a) either reasonably sophisticated, involving technical expertise such as programming for special uses and the like, or (b) "routine map-making", e.g., choropleth maps that are rarely linked to spatial analysis beyond subjective visual impression of map patterns.  The "routine map-making" exercise sometimes includes statistical analysis, but generally as a separate, rather than GIS-linked, endeavor.

Indeed, the use of even simple GIS/SA approaches would be a step towards resolving the aforementioned bifurcation.  Examples of such approaches might include the use of overlay mapping, calculating the degree of correspondence between two distributions, isorithmic mapping that generalizes spatial pattern, residuals from regression, or spatially varying parameters (termed, I believe, "spatially-weighted regression" in current literature).

This is a critical observation.  Even simple combinations of GIS and spatial analysis are not being exploited and/or widely used.  The sense of a "missed opportunity" is very obvious.

As noted, readily accessible GIS/SA software is critical to reversing the current situation, as are teaching GIS/SA and using it in published research on substantive issues.

A related concern is sensitivity by other physical/social scientists to issues in the GIS/SA realm.  In general, not widely understood or appreciated are the effects on analytical outcomes of areal size (boundaries and areal units cross-sectionally and/or in terms of their change over time; the modifiable areal unit problem), spatial resolution of data, and the meaning of contextual analysis from a spatial frame of reference (done in a manner that passes as spatial analysis for many, but would not satisfy the criteria of a GIS/SA geographer).

(5) Who Uses GIS/SA in Scholarly Research that is Substantively Motivated?

From my knowledge base, the use GIS/SA in scholarly research that is dominated by a substantive concern (and I emphasize scholarly rather than applied research, a certain brand of scholarly research at that) is more prevalent outside of geography; e.g., in anthropology, demography, epidemiology, sociology.  The basis for this observation is simply scanning major geography and regional science journals of general interest relative to parallel journals of other disciplines; also from reviewing research proposals of many disciplines.

This is of critical importance.  Disturbing in its own right is the point that GIS/SA might be used, and appreciated, more widely by substantively oriented professionals other than geographers-regional scientists.  In addition, however, I see evidence that GIS is increasingly being disassociated from Geography; that others take it as simply another tool, and the connection to Geography per se is diminished, if not lost.

On its face, the present balance exemplifies some of the issues I've outlined earlier.  There is evidence, that the balance is shifting.  More centrally, the point remains that we need to carry the GIS/SA revolution forward, stimulate its wide and informed use among geographers and others, and continue to gain credit and recognition for this important tool, for the G in GIS

(6) What is the Problem, How Can It Be Remedied?

The responses to this question are embodied in earlier comments.

First, much of the substantively motivated research being done is published in specialty rather than general journals.  Without neglecting the former, spreading an understanding of GIS/SA utility to a broad audience is highly important.

Second, we need readily accessible, user-friendly software that embodies the GIS/SA interface, including facilitated coupling with widely-used software such as ArcView.  Moves in this direction are underway, as represented by software such as SPACESTAT and S-PLUS, mentioned earlier. It also is possible that GIS/SA laboratories will facilitate moving this interface forward, and that software companies will include greater spatial analytic properties in their product.  As an example, the latest version of ArcView allows isorithmic mapping; but obviously, the GIS/SA interface must go well beyond that.

Third, we need courses, instructional direction, and workshops that focus specifically on the GIS/SA interface.  As noted, our course structure tends to emphasize GIS and SA separately.  Integration is needed.  The one workshop of which I'm aware (University of Michigan) is focussed on social science overall, draws few geographers, but in consonance with the message here, I think it tends to draw end-users rather than methodologists per se.

Fourth, we need substantive research, carried out by non-specialists in GIS/SA (or specialists), that provide a guide, "demonstration effect", and inspiration for others.  Ultimately, the success of revolution is its widespread use -- by day-to-day researchers who constitute the majority of Geography and our various disciplines.  Technical research and training professionals with technical expertise remains essential, but equally essential is a base of researchers that appreciates, motivates, and uses the product of the specialists' work.



 
 

Curriculum Vitae

http://www.geography.ohio-state.edu/faculty/brown/



 
 

Address

Lawrence A. Brown
Department of Geography, The Ohio State University
1036 Derby Hall, 154 North Oval Mall, Columbus, Oh 43210

Email: labrown@geography.ohio-state.edu


Go back to list