Quantitative Revolution

The term quantitative revolution describes attempts in the late 1950s and early 1960s to introduce mathematical and statistical methods into research and teaching in academic geography. Prior to this, academic geography had been concerned mainly with describing the characteristics of regions of the earth's surface using maps and text, emphasizing the unique characteristics of these regions. The origins of the term are obscure, but it was popularized in a paper by Ian Burton, in which he made two assertions. First, even in 1963, he was prepared to declare that the “revolution” was over, and, second, he claimed that the primary drive of the revolutionaries was toward well-founded, essentially positivist scientific theory. With the benefit of hindsight, both were only partly true.

First, almost all commentators agree that the revolution in both the United States and the United Kingdom continued throughout the later 1960s and well into the 1970s to become the mainstream geography of most of the 1980s. Its impact on the geography studied in continental Europe and much of the rest of the planet was minimal until almost the end of the 1970s. Although much has been made of its origins in the very talented group of graduate students that assembled at the University of Washington in Seattle during the late 1950s and of a supposed “Cambridge-Bristol” axis of diffusion in the United Kingdom, hindsight shows that the changes weren't quite as revolutionary as the protagonists, both for and against, claimed. Academic geography already had some strongly quantitative branches of which climatology, land survey, thematic mapping, map projections, and some parts of geomorphology are obvious examples. As a “revolution,” it was a long, drawn-out affair, and in no sense did it completely displace earlier concerns and approaches in the subject.

Second, although we need to be careful about the use of the term theory, the quantitative geography developed by most of its practitioners was essentially empirical and statistical in nature. As pointed out by Taylor and Johnston in their essay “Geographic Information Systems and Geography,” the term covers a very catholic set of approaches, the dominant one being the collection and analysis of data using standard aspatial statistical techniques. Given the complexity of much of the real world, it is hardly surprising that the data sets used were often highly multivariate in nature and were analyzed using, for example, multiple linear regression, principal components/factor analysis, and classificatory methods from numerical taxonomy. Hindsight has shown that given the special characteristics of spatial data, many of these methods were inappropriate. With perhaps a [Page 360]half-dozen exceptions, very few of the methods used were explicitly spatial in nature or have become part of the standard toolkit within GIS. Given the complexity of the calculations, it is also hardly surprising that the revolution, if that is what it was, had as an almost necessary precondition the increasing availability of the easily programmed, “mainframe” computers being introduced at much the same time.

It is the more innovative, explicitly spatial quantitative work undertaken by the so-called revolutionaries that has become embedded in what is now known as geographic information science and justifies the assertion by many authors that the revolution was a primary antecedent of geographic information science. First, often in collaboration, workers in academic geography and statistics, such as Andrew Cliff, Peter Diggle, Keith Ord, and Brian Ripley, produced methods for describing and testing for what can loosely be called pattern in spatial data.

...