Why Are Geographic Information Systems Hard to Use?

Carol Traynor and Marian G. Williams

Computer Science Department
University of Massachusetts Lowell
Lowell, MA 01854 USA
ctraynor@cs.uml.edu (+1 508 934-3385)
williams.chi@xerox.com (+1 508 934-3628)

Abstract

Geographic Information System (GIS) software evolved out of the fields of geography, cartography, and database management. As a result, off-the-shelf GIS software requires the user to have or to acquire considerable knowledge of these fields. Navigation through the interfaces of most off-the-shelf GIS software is difficult because they support a system architecture view, rather than a view of the user's work. These problems are compounded for users with little computing experience. In many workplaces, a single technical user becomes the local GIS expert, and acts as a surrogate for other users who have neither the expertise to use the software nor the resources to acquire that expertise. In this paper, we summarize our analysis of what makes GIS so hard to use, and describe our research directions toward designing effective GIS software for non-specialist users.

Keywords:

geographic information systems, partici- patory design, task analysis.

Introduction

A geographic information system is software for the acquisition, storage, analysis and display of geographic data. It combines cartographic tools with a data base management system.

Off-the-shelf geographic information system software is hard to use unless you have sufficient knowledge of geography, cartography, and database management systems; are computer-literate; and invest sufficient time to become accustomed to an interface that reflects the system architecture. It is best suited for use by specialists, or for people who can afford to make the investment of time and effort to become specialists.

We became interested in Geographic Information Systems (GIS) through a project in which some social scientists at our university are attempting to develop a GIS application for use by residents of the inner-city neighborhoods of Lowell, MA. The application will contain data about demographics, schools, crime, hospital emergency use, toxic waste sites, etc. The goal is to empower residents to lobby more effectively for change in their neighborhoods.

It became clear early in the project that, despite consultations with a GIS expert [3], the social scientists were finding the software very difficult to use. It also became apparent that there is little hope that community residents with no computing experience will be able to use the software.

Two additional projects have given us added insight into the difficulties of use of GIS. We have developed materials for introducing GIS to public school teachers who think they want to incorporate mapware into their curricula [4], and we have asked computer science graduate students to use various off-the-shelf GIS software packages as part of their graduate HCI classwork.

We have to date used, and have watched the various users mentioned above attempt to use, seven prominent off-the- shelf GIS software packages. Two of the GIS packages are Macintosh-based; five are PC-based. Five have mouse- driven graphical user interfaces (GUIs), one has a command line interface, and one requires both mouse and keyboard to navigate through menus. We generalize from these seven packages in order to explore the issues that make software of this type hard to use by non-specialists.

According to a task analysis, performed by Knapp [2], of GIS use by specialists, GIS tasks use four primary "visual operators": locate, identify, compare, and associate. The user interfaces of the GIS packages we have looked at do not explicitly support these operations. Rather, users must navigate through levels of menus that reflect the architecture of the software. In order to become adept at using the GIS software, they must learn to map their tasks onto that architecture.

DOMAIN KNOWLEDGE AND NAVIGATION A non-specialist user approaching a GIS will expect to work with maps. But in the GIS software we have looked at, the word "map" may not appear in its expected sense. For example, in one of the packages, a map is opened by selecting Geographic from the File menu, then selecting Use from the Geographic submenu, and finally typing a filename into a dialog box. The word Map appears in the View menu, but is used to define a cartographic view, not to open a map. In another package, a map is opened by selecting Open Table from the File menu.

We have watched computer science graduate students spend as much as an hour trying to figure out how to open a map in one particular GUI-driven GIS, and give up in frustration. They do not have a cartographer's vocabulary (terms like view, overlay, and thematic layer), nor have they a cartographer's understanding of the components that make up a map and of the relationships between those components.

We have looked at how the GIS packages support Knapp's four visual operators (locate, identify, compare, and associate), and have found that users may have to translate these operations into complex series of steps, using terms and concepts from cartography and database management.

Accomplishing user tasks in the GIS software packages that we have looked at is not intuitive for non-specialists because the tasks have to be carried out using commands that reflect the software implementation and require knowledge from several technical domains. Thus non- specialist users must translate their tasks, rather than having the computer support their work.

GIS FOR NON-TECHNICAL USERS

In many workplaces, a single specialist becomes the local GIS expert, and acts as a surrogate for other users who have neither the expertise to use the software nor the resources to acquire that expertise [1]. We observe this phenomenon happening within our university, where the social scientists mentioned above have hired some computer science graduate students to become GIS specialists. These specialists become immersed in the terminology of GIS and in a way of thinking about information that the architecture imposes on them. Even when coached, these specialists have trouble communicating with the users they serve, because of the burden of translating between GIS-speak and the workplace terminology of the users.

In order to support the work of the non-technical users - and to free them from dependence on surrogate users - we are engaged in the participatory design of a new GIS. The users who are currently involved in the project are school teachers and social scientists. We expect to work with community activists from the Lowell, MA neighborhoods, as well. The interface of the GIS will directly support the users' tasks and will protect them from having to know about how data are stored and how maps are drawn.

Acknowledgments

This research is supported in part by U.S. Department of Education grant number R117E30237, Improving the Success Rate of Adapting Technology for the Classroom. We are indebted to J. Nicholas Buehler for his valuable help and advice.

References

1. Garson, G. David and Robert S. Biggs, Analytic Mapping and Geographic Databases, Sage University Paper series on Quantitative Applications in the Social Sciences, series no. 07-087 (Newbury Park, CA: Sage), 1992.
2. Knapp, Loey, "A Task Analysis Approach to the Visualization of Scientific Data," Ph.D. dissertation, University of Colorado, 1994.
3. Kuhn, Sarah, Charles Richardson, and Marian G. Williams, "Meeting of the Minds: The Challenges of Interdisciplinary and Inter-occupational Communication," in PDC ‘94: Proceedings of the Participatory Design Conference. R. Trigg, S.I. Anderson, and E.A. Dykstra-Erickson (Eds.). Chapel Hill, NC, USA, 27-28 October 1994 (Palo Alto, CA: CPSR), p. 147-148.
4. Traynor, Carol, James Pash, and Marian G. Williams, "Bringing Users into the Discussion of an Unfamiliar Technology," in PDC ‘94: Proceedings of the Participatory Design Conference. R. Trigg, S.I. Anderson, and E.A. Dykstra-Erickson (Eds.). Chapel Hill, NC, USA, 27-28 October 1994 (Palo Alto, CA: CPSR), p. 131.