Which of the following are not activities typically associated with geodemographers

Geodemographics is a means for studying the differences between neighborhoods at a local, regional, or national scale which is useful for the evaluation of social and commercial policies. From modest beginnings in the first half of the twentieth century, the construction of geodemographic systems has developed into a sophisticated apparatus for the multidimensional analysis of large data sets. Contemporary geodemographic classifications incorporate data from commercial, as well as government sources. Geodemographics can benefit users – for example, in healthcare, retail, education, and policing – by allowing resources to be allocated with greater efficiency or effectiveness. However the approach is not without its critics, who have objected on both ethical and methodological grounds. Important ethical questions concern the propriety of using data about individuals in potentially discriminatory ways, while the methods could be viewed as a static and homogeneous representation of a dynamic and heterogeneous population. In spite of these drawbacks, the article will argue that geodemographics is a topic of considerable academic and practical importance.

6 Conclusion

Geodemographics is particularly hype-ridden and the success of its applications limited, for example, to very modest ‘lifts’ in customer response by 5–10 percent in direct mail campaigns, even if this represents considerable savings in a large-scale operation. It is, nevertheless, the fastest growing segment of the information industry, presumably because of its promise of intimate and actionable knowledge about consumers. Geodemographics first reduces the subjectivity of individuals to objective and quantifiable characteristics, then reconstructs composite ‘soap opera’ identities (Goss 1995, p. 187), perhaps more consistent with marketers prejudices than social reality—hence one geodemographic system promises that ‘as you get to know [our cohorts], you'll discover these are all people you know: your friends, neighbors, relatives, and—most importantly—your customers’ (National Demographics and Lifestyles 1993, p. 2). There is a second, more sinister, self-fulfilling promise, however, in that the definition and targeting of consumers based on residential location may effect redlining of consumption, such that marketing campaigns and retail services are provided only to addresses—with known names, inferred names, or merely ‘Current Resident’—with the ‘right’ statistical and behavioral profiles. This is not to argue that there are not cleavages in consumption based on class and residential location, but if, for example, magazine issues are differentiated in terms of advertising content for subscribers in different ZIP codes, as some already are in the US, consumption of these goods and services in turn will reproduce, and perhaps actually produce, further segmentation and segregation of society. Such is the instrumental reason behind the hype of geodemographics.

Clustering Method and Cluster Frequencies

By and large most geodemographic classifications are created with k-means clustering either entirely to create large aggregations that are successively split into smaller subsets to build a hierarchy, or are clustered using this method at the finest level, and aggregated up using a hierarchical method such as Wards. Others have implemented alternative approaches such as k-medoid clustering; however, in evaluation, the results across all approaches and methods are broadly similar, and make limited substantive difference to the end results. One area of contention when building a geodemographic classification relates to the selection of an appropriate number of clusters that describes both the given problem and are appropriate given the input data selected. There are number of heuristics used within this context, however, it is most common to look at changes in total within sum of squares (how well the clusters fit the data) for different numbers of clusters. The challenge is balancing frequency and detail, with interpretability and saliency between clusters. For operational classifications where there is stakeholder involvement in the classification process, this is usually a key area for discussion and comparative evaluation of different scenarios.

General and Bespoke Geodemographic Classification

Most geodemographic classifications are created for general purposes; that is, to describe the overarching characteristics of the population and those places within which they live. Such classifications are not optimized for any particular purpose, other than aiming to reflect those salient groups deemed to make up society; and within the contemporary period, they are typically created for national extents. Such orientation lends itself to exploring urban structure by mapping out aggregate national patterns of cluster differentiation, but also, by examining differences in rates of consumption or behavior between residents of constituent clusters, to explore at how such patterns vary locally, regionally or nationally.

An alternative approach observed within the commercial and academic sector has been to build geodemographic classifications to address a particular problem through inclusion of measures related to the phenomenon being studied (health, crime, education, etc.); or, focusing a classification to a specific geographic extent (e.g., a metropolitan region). Within the commercial sector there is now a proliferation of commercial classifications targeted at particular sectors including automotive, finance, and the public sector more generally.

An example of a bespoke geodemographic classification is shown in Fig. 2. This classification was specified for a single region of Greater London but also drew on data describing workplace structure and activity.

Figure 2. The London Workplace Zone classification showing the geography of workers and their contexts within the Greater London Authority region.

Geodemographics

Although the term ‘geodemographics’ was first used by geographers more than a century ago to classify neighborhoods within cities, its contemporary usage can be attributed to its use within business demography to refer to the analysis of demographic data, usually using relatively small spatial units, for the purposes of market analysis and related purposes. The term and approach have more recently been used in other contexts, especially for social research and the provision of public services. One of the central approaches of geodemographics is the clustering of statistically similar neighborhoods or other areas. Census data are usually central to these approaches since geodemographics demands information at a detailed spatial scale and often involves a number of variables. Census data are then often supplemented by data from other sources such as market surveys, records of housing sales, electoral rolls, and so on.

Geodemographics

Although the term “geodemographics” was first used by geographers more than a century ago to classify neighborhoods within cities, its contemporary usage can be attributed to its use within business demography to refer to the analysis of demographic data, usually using relatively small spatial units, for the purposes of market analysis and related purposes. The term and approach have more recently been used in other contexts, especially for social research and the provision of public services. One of the central approaches of geodemographics is the clustering of statistically similar neighborhoods or other areas. Census data are usually central to these approaches since geodemographics demands information at a detailed spatial scale and often involves a number of variables. Census data are then often supplemented by data from other sources such as market surveys, records of housing sales, electoral rolls, and so on.

4.3 Segmentation Methods

Segmentation can be carried out based on geographic, demographic, geodemographic, behavioral, lifestyle, and psychological characteristics (De Saez, 2002, pp. 118–123). Characteristics such as gender, age, race, occupation, religion, and education are used in demographic segmentation (De Saez, 2002). Geographic segmentation is a simpler segmentation based on regions and locality. Behavioral segmentation takes into account usage statistics, for example, the number, type and branch location of borrowing to differentiate the market (Millsap, 2011). There are many segmentation methods, and more than one may be required to achieving distinctive segments. Common methods include:

Geographic segmentation refers to dividing a market into different geographical units;

Demographic segmentation means dividing the market into segments based on variables such as age, life-cycle stage, gender, income, occupation, education, religion, ethnicity, and generation;

Psychographic segmentation is dividing a market into different segments based on social class, lifestyle, or personality characteristics; and

Behavioral segmentation refers to dividing a market into segments based on consumer knowledge, attitude, uses, or responses to a product (Kotler & Armstrong, 2014, pp. 193–198; Kotler and Keller, 2009, pp. 213–226).

Geodemographics

A major application of classification–regionalization procedures during the past two decades has been in an area commonly known as geodemographics. This developed from factorial ecology. The procedures and algorithms involved were adopted by market research firms to identify areas with common socioeconomic and demographic characteristics that could be used for targeting in various sales campaigns. The procedures were enhanced after the development of geographical information systems allowed the census information to be linked with other data sets (many of them proprietary) based on consumer and other surveys; with these, it is possible to update data sets regularly rather than relying on census data, which may be 10 or more years old. In addition, in many countries the address files associated with the postal and electoral systems are also available for purchase, which can also be used to update address files for targeting.

The types identified using such procedures classify areas according to their residents' lifestyles based on information relating to such characteristics as newspaper readership, types of TV programs watched, and frequency of purchase of various products as well as indicators derived from censuses (such as age and socioeconomic status): they are sometimes called lifestyle databases. these allow firms to target their marketing at certain types of areas, and thus customers, thereby either avoiding areas whose residents are unlikely to purchase their products or identifying areas for potential expansion of sales. In addition, such databases may be used by other bodies, such as political parties seeking to identify target groups of voters for particular policies.

Whither Social Measurement in Urban Studies?

Beyond the academy, there continues to be very strong interest in social measurement of urban socioeconomic distributions, and many interesting developments of technique and application are underway. One important development has been the recognition that richer depictions of urban lifestyles can be assembled from diverse and under-used digital sources that are available at a range of spatial scales. These include the limited range of attributes that can be gathered for individuals from address registers, to georeferenced vehicle registrations, social surveys, shopping surveys, and guarantee card returns. Increasing numbers of commercial databases concerning lifestyles are now routinely used in commerce: many enumerate salient characteristics of tens of millions of individuals. Although they are rarely collected to the same exacting standards to which conventional public sector data sets are collected, they are updated continuously and provide very detailed snapshots of the diverse lifestyles that are often to be found in quite small urban areas. Some enlightened private sector data providers have been prepared to deposit these disaggregate data in research archives (subject to confidentiality constraints). Although these vastly enrich the potential content of social classifications, their use raises a number of profound and possibly frustrating issues of coverage and representativeness, not least because commercial organizations are unlikely to be motivated toward assembling detailed data inventories on the “have nots” of society. Companies that use lifestyle lists to identify potential mail-drop targets are unlikely to select households with low incomes. For this reason, most lifestyle operators have decided that it is more cost effective to target the blanket door drops of questionnaires to postcode sectors with higher rather than lower levels of affluence. However, coarse administrative zones (such as UK postcode sectors) are the lowest areal units that can be leafleted by distributors, and this is likely to improve the representativeness of response. There is a clear need to cross-validate small area lifestyle measures with respect to external data sources.

These issues illustrates a number of the tensions in current social research: digital data capture is now routine, but scientific surveys conducted to rigorous standards account for a diminished real share of available data; the scale and pace of change to urban systems now makes the decennial snapshots of censuses increasingly irrelevant to policy needs; and the fission of lifestyles among urban populations that are increasingly heterogeneous at fine scales of granularity makes the limited content (attribute base) of censuses increasingly limiting in analysis of urban systems. Taken together, this argues for the need to generate scientific findings in real time, and with frequent update cycles; the need for interdisciplinary approaches to urban studies that blend together salient indicator variables into comprehensive indicators; and the need for public–private partnerships in order to unlock the potential of the richest, most relevant, and most recent data.

In relating the success of social measurement through geodemographic applications in business, it is important to note some differences in the measurement goals of different academics and private sector practitioners. In general terms, it is worth distinguishing between two rather different applications. Direct marketers, who communicate with individuals rather than serve areas, desire measures that are optimally predictive at the person or household level, but are not materially concerned whether there is any geographically systematic error in this estimate. There are no inferential errors generated in one-to-one marketing applications. By contrast, retailers and other organizations that serve areas, rather than individuals, are not particularly concerned whether their social measures are accurate at the household level. They, like the academic or policy analyst, are more concerned that whatever inferential errors there may be are not systematic at the area level. Thus, the best social measurements for direct marketers are not necessarily the best measures for analysts concerned with geographic catchments.

Today's diminished academic interest in social measurement of urban systems is a pity for a number of reasons. In conceptual terms, the experience of factorial ecology did bring general recognition of the ways in which choice of classification method, choice of variables, and to some extent choice of data source would determine the outcome of the classification. Today, similar sensitivity to context is recognized in the measurement of local or regional effects. Second, in measurement terms, more data are collected about more aspects of our individual lifestyles than at any point in the past, through routine interactions between humans and machines. Enlightened approaches to public data access (especially through online portals) make wide dissemination of socioeconomic data a reality and the creation of general-purpose and bespoke data systems straightforward. Geodemographic systems based on socioeconomic framework data can be successfully fused to census sources to provide richer depictions of lifestyles. And third, in analysis terms, the toolkit of spatial analysis and GIS now make it easier than ever before to match diverse data sources and accommodate the uncertainties created by scale and aggregation effects.

Developments in computation, technique, and data analysis continue to offer incremental improvements in the ways that geodemographic representations are specified, estimated, and tested, but it is correct to suggest that it is repeat purchases of a core tried and tested technology that ensure retention of the approach as a mainstay of contemporary urban studies. Hitherto, the overwhelming majority of geodemographic applications has concerned tactical and strategic decision making in private sector applications (specifically retailing), and it is probably true to say that the clearest indicator of “success” is the way in which improvements in targeting of goods and service offerings improve measured profitability. One of the interesting challenges of the coming years will entail use of these techniques in public service applications, given the pressures to demonstrate value for money in targeting public funds according to local needs. It is not possible in an article of this length to examine the various caveats to the geodemographic approach, but issues of the content and coverage of the data sources that are used to create and update geodemographic profiles are certainly likely to become important in developing and extending the realm of geodemographic applications. The approach has very important contributions to make to the developing rationalities, performance metrics, and change measures in the developing public policy debate. It is of strategic importance that academics and policymakers engage in these important measurement issues, and do not simply become passive consumers of geodemographic systems. In this context, it is important to return to the themes of conception, measurement, and analysis. In the early days of urban studies, the issue of empirical verification of ecological analogy was seen as a crucial issue. Much of the recent history of urban studies has also posited concepts and processes that can only rarely and unsystematically be observed. The geodemographic approach, by contrast, entails a return to the mosaic metaphor of urban structure, and provides robust, transparent, and disaggregate observations of what is going on in urban systems. In conceptual terms, it is founded on the basic theoretical premise that “birds of a feather flock together.” This is by no means a trivial concept, whether viewed in the context of genetic selection and mapping of the human genome, or in the simulation of city evolution as the outcome of cellular interactions across a range of geographic scales. It has been suggested that the patterning of urban social areas may express more than the outcome of (unmeasurable) economic and social processes and, at a conceptual level, that there may be much that can be developed from the success in measuring urban phenomena through geodemographics. It is not just the urban ecology of the Chicago school that might be revitalized: neighborhood classification could also develop the ideas of Shevky and Bell in terms of classifying urban areas according to the range and intensity of social relations, differentiation of industrial function, and increasing complexity of the segregation of society.

Geographical Distribution

The telephone has influenced substantially the distribution of people across the physical landscape. On the one hand, the telephone has allowed the massive vertical downtown urban center to survive then thrive. Without the telephone, the skyscraper would have been most difficult to build and manage. Further, it would be difficult for the structure to maintain its usefulness to its denizens without the telephone.

Some claim that the telephone has enabled one particular form of geodemographic dispersion, the suburb. However, detailed studies of transportation system development indicate that the telephone was not an important factor in this remarkable internal migration. Still, the telephone does allow many in rural locations to participate in business and social relations in a viable way, which without the telephone would not be possible. At the same time, those in extremely low-density areas have persistent difficulty getting even minimal telephone service; the situation is exacerbated by the fact that this populace is generally poor and remote from economic opportunities. This, unfortunately, is a social problem that will endure for decades. The telephone, though, has allowed those who choose isolation – and such people are often drawn from the middle and upper strata of society – to enjoy it without having to forego ready communication. Some seek to ‘get away from it all’ but still want employment as consultants, or at least to stay in convenient contact with friends, family, and emergency services.

Current Trends in Location Analysis

Location analysis research and applications of today differ from that of 30, 20, or even 10 years ago. One of the major developments is the recent widespread availability of highly functional, low-cost GIS and related computing technology. Many location problems can be approached in a sophisticated fashion from the confines of a desktop computer. Related to both the growth of GIS and spatial modeling, there has been continued development of the tools and techniques used to conduct market area analysis. Some of these tools take advantage of the increased power afforded by GIS to accurately represent real-world environments including population demographics, transportation systems, and facility locations. Lastly, the growth and deployment of the commercial Internet has profoundly influenced the way location analyses are currently practiced.

Trade Area Analysis, Geodemographics, and GIS-T

Location analysts frequently need to determine the extent of trade areas, identify the characteristics of people living within those trade areas, and relate this information to key variables about the facilities, such as their sales, production, etc. Trade area analyses have become more sophisticated as the range of computer-based tools available to analysts has improved in quantity and quality. The ability of GIS to represent detailed trade area geographies and measure the characteristics of people within them to the household level makes it an important tool for businesses wanting analyze their markets.

Two principal developments have helped facilitate more robust trade area analysis. First, there has been an increase in the amount of detailed spatial data available about consumers, their characteristics, and their product preferences. Some of these data are obtainable from public sources, but many private data vendors sell geodemographic data. Much of these data are used for market segmentation purposes where the analyst seeks to determine if there is a pool of demand for a particular product based on the area’s demographics. Geodemographic vendors build profiles of people in areas that account for their income characteristics, cultural affluence, political leanings, and many other dimensions.

Second, specialized goegraphic information systems for transportation, referred to as GIS-T, have contributed to more insightful market area analysis. GIS-Ts are particularly adept at managing and analyzing geographic data that involve transportation features. Because road networks are important for gaining realistic estimates of drive times or distances between stores and customers, as well as good distance measures between locations, GIS-T has played a growing role in location analysis. Figure 3 shows an example of a travel distance contour as constructed in a GIS-T. It shows travel distance contours along a network from a selected intersection (1, 2, 3, 4, and 5 miles).

Figure 3. Travel distance bands of 1–5 miles from a selected intersection in Leon County, FL.