Longitudinal Analysis of Age- and Gender-Specific Migration Patterns in England and Wales: A GIS-Based Approach

Many geographers have argued for the need to incorporate change over time into their analyses (see, for example, Haggett 1965; Hägerstrand 1970; Thrift 1977; Marsh et al. 1988); however, changes to administrative boundaries often mean that demographic statistics collected at two different dates cannot be directly compared. This has made it very difficult to study longitudinal change without resorting to undesirable levels of aggregation, typically to county level in Britain or state level in the United States. This article describes a technique that makes significant advances toward eliminating this problem: a researcher using this technique can compare statistics by standardizing all relevant data on a single set of administrative units. My article builds on the work of the Great Britain Historical GIS Project (Gregory and Southall 1998) and uses net migration as an example. The geographical information system (GIS) is not yet complete, so the article focuses on the methodological issues. These issues could be applied to a wide range of problems in historical geography. More broadly, it is hoped that the article will give some idea of the potential for using GIS to analyze spatially referenced data in the context of social science history.

The article uses the county of Gloucestershire in western England as an example. The boundaries of the county's primary administrative units over the period from 1881 to 1931 are shown in Figure 1. Between 1881 and 1911 Registration Districts were the primary units used, for example, for publishing census and vital registration data. Their boundaries changed significantly over the 30 years, especially around Bristol in the southern part of the county. After 1911 a new administrative geography, based on Local Government Districts, was used. This geography contains many new areas because towns were usually set up as separate districts, and many areas were redefined and renamed. It is also apparent that the shape of the county changes at all three dates. Between 1911 and 1931 this was because the Registration County used in 1911 was replaced by the Administrative County; between 1881 and 1911 it was because of the expansion of Bristol.

For a researcher interested in decadal net migration, these changes cause major problems: the place called Bristol was a different area at all three dates; the place called Westbury on Severn in 1911 was four separate areas in 1931; Barton Regis was abolished, but its data cannot easily be reallocated to another district through aggregation; and there are many minor boundary changes, such as the Cheltenham and Tewkesbury boundary, between 1881 and 1911. This essay attempts to solve these problems by using GIS to interpolate data from all dates onto a single administrative geography, thus allowing direct longitudinal comparison. A major aim of the methodology is to preserve as much of the original detail contained in the source data as possible. To this end, data are interpolated onto Registration Districts as configured in 1911. Although Gloucestershire is used as an example, once the GIS is complete it will be possible to calculate decadal net migration rates for the whole of England and Wales from the 1850s to the present. It will also be possible to interpolate onto a different standardized geography, for example, modern districts or parliamentary constituencies.

Figure 1.

Administrative boundaries in Gloucestershire: 1881, 1911, and 1931. Dotted lines join detached portions of districts to their main area.

Academic geography is still searching for a niche into which GIS can be inserted. At one extreme, enthusiasts such as Stan Openshaw argue that GIS will revitalize and reunite geography by creating a cohesive, scientific framework around which all other work can be based (see, in particular, Openshaw 1991b, 1997). At the other, there are those who are fundamentally opposed to GIS, arguing that it marks a return to "the very worst sort of positivism" (Taylor 1990: 211) and that it lacks a strong epistemology and any treatment of ethical, economic, or political issues (see, in particular, Pickles 1995a, 1995b; Curry 1995).

The field where GIS...