Geography of Diffusion Research Paper

While geographers have studied diffusion through most of the twentieth century, the focus has shifted over time, often with marked departures from prevailing norms. Consequently, diffusion studies by cultural geographers earlier in the century may be hardly recognizable as similar in genre to those at the end of the century, focusing on entrepreneurial agents in diffusing products or technologies.

2. The Cultural Geography Tradition

Diffusion has been a subject of geographic enquiry for much of the twentieth century. The spread of a phenomenon, idea, or technique through a population or region incorporates basic geographic elements of distance, direction, and spatial variation. Within geography, diffusion studies originated in cultural geography, whose research goal is:

… to identify environmental features characteristic of a given culture and…discover what role human action…has played in creating and maintaining given geographic features (Wagner and Mikesell 1962, p. 1)

The cultural geography tradition dominated pre-World War II scholarship and focused on change in indigenous cultures, often centuries ago. Culture and the human or cultural landscape are the main concern, and diffusion studies are relevant in clarifying questions concerning culture origins, the spatial configuration of culture areas, and culture-related or culture-influenced features on the earth’s surface. Carl Sauer was the central figure in this approach, which tends to be closely aligned with and complementary to some concerns of anthropology (Mikesell 1967, 1978).

Within this framework, many studies aim to assess the relative importance of diffusion vs. simultaneous invention in development of cultural landscapes. They employ an empirical, inductive approach with emphasis on landscape development in prehistory and/or over considerable periods of time. Often investigation considers a single phenomenon, seen as a culture trait, and explains its observed spatial distribution as the outcome of processes unfolding over space and time. A typical procedure is to examine a phenomenon’s spatial distribution at different times or its frequency of occurrence in different locations at a single time. Inferences are drawn regarding origins (or cultural hearths), means, and routes of dispersal of both the observed phenomenon (or culture trait) and ideas– techniques–culture for which it is a surrogate. Attention also is given to cultural and environmental conditions necessary for adoption of particular phenomena (Bobek 1962) and to macroscale, big picture phenomena, such as the domestication of plants and animals, their origins, routes of dispersal, and destinations (Sauer 1952).

In stressing the role of diffusion in development of the human landscape, many cultural geography studies proceed further to consider indirect effects of the culture trait innovation on physical and cultural environments. A basic notion is that the diffusing item is both a stimulus to new innovation and itself subject to modification as it spreads. The relation between diffusion, the item being diffused, and the human landscape is therefore complex and subject to continual change.

3. The Hagerstrand Tradition

Sauer’s counterpart in later diffusion research is Torsten Hagerstrand who, in the early 1950s, initiated a shift towards the generative processes underlying diffusion and modern-day products and practices. Since this focus is consistent with Sauer’s conceptualization, why was process less prominent in earlier research? One reason may be that identification of processes controlling locational change is less feasible in the sparsely documented situations studied by cultural geographers. Second, Hagerstrand’s main impact was on researchers in economic and urban geography who were primarily concerned with location and locational processes rather than landscape. Illustrative is Haggett’s (1965, pp. 10–13) classification of research under the Sauer school as landscapist or ecologist and research under the Hagerstrand school as locationalist.

Hagerstrand’s research (1952, 1967) focuses on diffusion of contemporary innovations such as the automobile and bovine tuberculosis control. His first study, The Propagation of Innovation Waves (1952), continues the cultural geography tradition by its use of an inductive approach and concern with diffusion pattern rather than process. At a somewhat later date (1953), however, Hagerstrand observed that

… spatial order in the adoption of innovations is … so striking that it is tempting…to create theoretical models which simulate the process and make predictions achievable (Berry and Marble 1968, p. 370)

This statement indicates a change from description and inductive generalization to a more deductive approach focusing on generative processes. The work marking this shift is Hagerstrand’s Ph.D. dissertation of 1953, subsequently translated into English as Innovation Diffusion as a Spatial Process (1967). Hagerstrand made three contributions that continue to play a role in diffusion research.

3.1 Empirical Regularities In Diffusion

Three regularities characterize diffusion. Over time, a graph of the cumulative level of adoption is expected to approximate an S-shape that rises with the passage of time. In an urban system, diffusion is expected to proceed from larger to smaller centers, a regularity termed the hierarchy effect. Within the hinterland of an urban center, diffusion is expected to proceed in a wave-like fashion outward from the urban center, first hitting nearby rather than farther-away locations, and a similar pattern is expected in diffusion among rural populations. This regularity is termed the neighborhood or contagion effect.

Hagerstrand (1952) also provides a framework that connects these regularities. For diffusion in an urban system, this posits three phases: a primary stage, during which initial diffusion centers are established; a diffusion stage, during which neighborhood-type diffusion occurs in areas near diffusion centers and secondary diffusion centers are established in lower-order urban places; and a condensing or saturation stage, during which filling-in occurs and diffusion eventually ceases.

This framework thus illustrates the complementarity of hierarchy and neighborhood effects and their temporal positioning within diffusion. Also, if locales of adopters were tabulated, higher-order urban places characterize the left tail of the S-curve of adoption; lower-order urban places characterize the middle, sharply rising portion of the S-curve, along with rural hinterlands in proximity to urban centers; and remote rural hinterlands characterize the S-curve’s right tail.

3.2 Hagerstrand’s Conceptualization Of The Diffusion Process

Hagerstrand’s conceptualization of innovation diffusion across the landscape holds that adoption is primarily the outcome of a learning or communications process. Factors related to the effective flow of information are most critical and, therefore, a fundamental step in examining diffusion is identifying spatial characteristics of information flows and resistance to adoption.

In Hagerstrand’s conceptualization, information originates from the mass media or past adopters. The destination of personal messages depends on a sender’s network of social communications (interpersonal contacts), and configuration of this network reflects various social and terrestrial barriers that impede, divert and/or channel communications. Particular attention is given to terrestrial barriers such as lakes, forests, difficult terrain, and geographical distance separating potential communicants.

The conceptualization further recognizes that resistance levels differ among individuals, and that higher levels of resistance require more information for adoption to occur. This leads to a dichotomy of resistance as either the result of values which are inconsistent with adoption (social resistance) or practicalities which make adoption difficult or impossible (economic resistance).

To address diffusion at different geographical scales, a hierarchy of networks of social communication is posited. One network might operate locally and another regionally. The first would control diffusion among farmers in a single locale, whereas diffusion at the central-place level of aggregation might utilize a regional network of persons in different urban areas that communicate with one another. Superimposition of these two levels provides a comprehensive picture of diffusion in the area represented.

In summary, Hagerstrand’s conceptualization proposes transformation of a population from a low proportion of adopters to one with a high proportion, by means of information dissemination through media and interpersonal contact. Spatially, the principal mechanisms of transformation are networks of social communication characterized by biases and distortions that, in turn, effect and are reflected in spatial patterns of innovation diffusion.

3.3 Hagerstrand’s Methodological Framework

To implement his conceptualization, Hagerstrand used Monte Carlo simulation. Adoption takes place after a specified number of messages are received, the specified number varying according to an individual’s resistance to adoption, which is randomly assigned. Time is treated in discrete units of equal interval, and each past adopter sends a message in every time interval. The destination of messages depends on the probability of contact between tellers and potential receivers. This probability is a function of distance between them unless a water or forest barrier intervenes. A distance decay function is employed to compute the probability of interaction between each of 25 cells in a 5-by-5 cell grid and the center cell of that grid. The resulting grid of probabilities, termed the mean information field (MIF), is employed with a map gridded in a similar manner, but with many more than 25 cells. The MIF is centered over a map cell containing an adopter of the innovation at time t; a random number is then drawn to determine which of the 25 surrounding map cells receives a message to adopt in time t l. If the message passes over a water or forest barrier, it is either nullified or a new random number is drawn, depending on the level of obstruction. This procedure is repeated for every adopter in every time interval. If a previous adopter receives a message, the message is lost, that is, it has no effect and is not repeated. If a message is received by a nonadopter who, owing to resistance level, requires more than one message before adoption, the message is stored until sufficient messages for adoption are accumulated.

This model mechanism is crude by today’s standards of computational capability, but in the 1950s it was highly innovative. Accordingly, Hagerstrand’s diffusion model played an enormous role in Geography’s quantitative revolution of the 1960s and 1970s, as both an example and standard of excellence (Berry and Marble 1968). This observation also applies to social science overall; the model is included, for example, in books such as Coleman’s (1964) Introduction to Mathematical Sociology.

4. The Market And Infrastructure Perspective

Hagerstrand’s formulation, and those of the many other social scientists studying diffusion from the 1950s onward, took an atomistic view of society as individuals with free will, any of whom could adopt an innovation. Thus, the onus of adoption was on individuals, who were classified on a scale ranging from innovator to laggard, with the social meaning each term implies (Rogers 1995). This is termed the adoption perspective, focusing primarily on the demand side of innovation diffusion.

The market and infrastructure perspective focuses on supply or availability of the innovation (Brown 1981). Rather than assuming all have an equal opportunity to adopt, this is seen as purposely and egregiously unequal, so that adoption represents a choice within a constraint set; government, non-profit, and entrepreneurial institutions establish these constraints; and attention is given to the process by which innovations and conditions for adoption are made available to individuals, households or firms. Hence, focus shifts from the individual to the structure within which individuals operate, paralleling realignments in social science overall (Brown 1999). Nevertheless, the market and infrastructure perspective is put forth as complementary, not contradictory, to the adoption perspective. Broad outlines of diffusion are shaped by where (and to whom) innovations are (made) available, but within those outlines, more detailed variation in diffusion patterns reflects adoption behavior.

Central elements of this perspective include the propagator of the innovation, the entity behind its diffusion; diffusion agencies (outlets) through which innovation is distributed to the population at large; and diffusion strategy, which is designed to encourage adoption and includes decisions on agency location, pricing, advertising, population segments targeted by promotional efforts, etc. Hence, locations and temporal sequencing of diffusion agencies determine where and when an innovation will be available and provide the general outline of the diffusion pattern. Diffusion strategy further defines the pattern by creating different levels of access depending on a potential adopter’s economic, locational, demographic, and social characteristics. The final pattern occurs via adoption behavior, the last stage of the process. Interacting with these factors are infrastructures like transpotation and utility networks that are independent of the innovation’s propagator and diffusion agencies, but distinctly affect adoption.

Hence, the market and infrastructure perspective emphasizes the marketing aspects of diffusion, especially strategies for market penetration. Whether the propagator is profit or non-profit motivated, broad acceptance is a common goal, and a strategy will be designed to accomplish this, that is, to best penetrate the market. This also is true whether propagation involves centralized decision making, decentralized decision making with a coordinating propagator, or decentralized decision making without a coordinating propagator, although such distinctions affect procedures by which diffusion occurs. Similarly, diffusion agencies may be represented by salespeople or agricultural extension agents, but either should be guided by marketing strategy and, accordingly, make selective decisions, regarding which potential adopters will be contacted or cultivated, and to what degree. Considering the role of infrastructure, broadcast television stations were needed before television sets diffused. Today, cable and satellite services vary enormously from place to place, affecting innovation availability. Enhanced accessibility through communication or transportation improvements increases the market areas of diffusion agencies, and renders innovations more attainable to potential adopters. Also, for example, irrigation infrastructure is often critical to the success of high-yielding variety seeds in agriculture.

Finally, this framework calls into question hierarchy and neighborhood effect regularities, because relevant factors from a marketing perspective are often not arranged in that manner. The diffusion of cable television in Ohio, for example, occurred initially in small, more remote places where broadcast signals were weak or nonexistent, because cable was a means of enhancing television sales by appliance stores.

5. Technological Diffusion

Agricultural technology has been a major focus of diffusion studies. Beginning in the 1930s, and accelerating thereafter, rural sociologists and agricultural economists, working through agricultural experiment stations and extension agents, carried out an enormous number of studies focused on farm innovations, such as hybrid corn, fertilizers, silage methods, high-yielding varieties, and the like (Rogers 1995). Until recent industrial agriculture, the adopting unit was the farm household, thus joining this effort with diffusion research concerned with consumer innovations. Manufacturing firms that are well beyond household size, also adopt technological innovations, as do service and retail firms. This has been studied extensively, but rarely joined with the body of diffusion research described previously.

Rather than emphasizing communications, studies of firm adoption give more attention to characteristics of the firm and innovation itself. Also significant is the fact that innovation replaces an older technology with a newer one, highlighting comparative costs, benefits, and risks; factors such as relative profitability, in-vestment, sunk costs, and the like. In this regard, there is greater emphasis on innovation as a continuous process; one that might be tailor-made for a particular firm or firm setting, rather than as a singular, unchanging phenomenon; and one that involves a more intimate and ongoing relationship between the innovation’s supplier and user. Issues such as competition among firms and relative factor prices, favoring one technology over another, also play a role.

As these factors indicate, economic perspectives are important to investigating the diffusion of technological innovations among firms (Brown 1981). But spatial considerations enter in two ways. First, there has been a tendency to assume that larger firms in larger urban agglomerations are earlier adopters, giving rise to a self-reinforcing (cumulative and circular causation; Myrdal 1957) hierarchy-effect pattern of diffusion. In fact, earlier adopters are just as likely to be less-than-the-largest in size, which may correlate with aggressiveness and innovativeness of management. There also may be a spatial order to innovative change, such that newer economic regions are favored, leading to regional rotation or spatial succession; Silicon Valley provides an example (Malecki 1997). The second way spatial considerations come into play is that the study of technological innovation has become closely intertwined with regional development and regional economic change (Malecki 1997). Hence, the questions may not be inherently geographic in terms of process, but very geographic in outcome.

6. Development As A Diffusion Process

Seen as integral to economic development, diffusion of innovations in Third World settings has received much attention. Examples include improving agricultural production and living conditions; generating entrepreneurial activity and related employment; promoting family planning, and improving infrastructure. Research and policy have largely employed the adoption perspective, but examples of a market and infrastructure approach are also evident (Brown 1981, Rogers 1995).

More broadly, development itself has been treated as a diffusion process. One approach involves choosing an index of development, and tracing its spread across the landscape; this approach is similar methodologically to the culture trait studies of earlier diffusion research. Such indices include agricultural cooperatives, banking, communications (telegraph, telephone), educational facilities, electricity, employment patterns across economic sectors, medical services, police organizations, postal agencies, transportation modes (roads, rail), urbanization, and other artifacts of a market economy (Gould 1976, Pedersen 1975, Riddell 1976, Taaffe et al. 1963).

Frontier expansion also has been viewed in a diffusion framework (Hudson 1969). Substantively, these studies focus on aspects such as capital generation and control; migration, travel, and trade interactions; the role of particular economic sectors such as banking and wholesaling; and transportation linkages (Meyer 1980, Pred 1980)

Finally, formal models of development have a distinct diffusion dimension. Modernization, dual economy, core-periphery, growth center, and stages of growth frameworks put forth the notion that Third World settings are composed of traditional and modern elements, and that development consists of transforming traditional to modern. Dependency and related theories take a similar tack, albeit one that differs in emphasis. These models divide Third World landscapes into dynamic, propulsive, modern cores and traditional peripheries; development is represented by transformation and expansion of the modern; and this occurs through mechanisms such as information and capital flows, migration, human capital developments, and more generally, backwash, trickle-down, and imitation effects (Brown 1991).

7. Overall Observations

Diffusion comprises a long tradition of research in geography. From epoch to epoch in social science, it has changed focus and content. Geographical studies of diffusion also represent only one approach in that the topic is highly multi-disciplinary including, at least, contributions from anthropology, communications, economics, marketing, political science, psychology, and sociology. Accordingly, concepts used by geographic studies of diffusion may be found in research by other disciplines; for example, interpersonal interactions, mass media effects, utility maximization, and marketing strategies. In all these instances, as well as in purely geographic studies, the critical contributions of a spatial perspective to understanding the dynamics of diffusion are apparent.

Bibliography:

1. Berry B J L, Marble D F (eds.) 1968 Spatial Analysis: A Reader in Statistical Geography. Prentice Hall, Englewood Cliffs, NJ
2. Bobek H 1962 The main stages in socio-economic evolution from a geographical point of view. In: Wagner R L, Mikesell M W (eds.) Readings in Cultural Geography. University of Chicago Press, Chicago, pp. 218–47
3. Brown L A 1981 Innovation Diffusion: A New Perspective. Methuen, London
4. Brown L A 1991 Place, Migration, and Development in The Third World: An Alternative View. Routledge, London
5. Brown L A 1999 Change, continuity, and the pursuit of geographic understanding: Presidential address. Annals of the Association of American Geographers 89: 1–25
6. Coleman J S 1964 Introduction to Mathematical Sociology. Free Press, New York
7. Gould P R 1969 Spatial Diff Resource Papers Series, Association of American Geographers, Washington, DC
8. Gould P R 1976 Tanzania 1920–63: The spatial impress of the modernization process. In: Knight C J, Newman J L (eds.) Contemporary Africa: Geography and Change. Prentice-Hall, Englewood Cliffs, NJ, pp. 423–37
9. Hagerstrand T 1952 The Propagation of Innovation Waves. Lund Studies in Geography, Gleerup, Lund, Sweden
10. Hagerstrand T 1967 Innovation Diffusion as a Spatial Process. University of Chicago Press, Chicago (trans. of Innovations for loppet ur Korologisk Synpunkt. Gleerup, Lund, Sweden, 1953)
11. Haggett P 1965 Locational Analysis in Human Geography. Edward Arnold, London
12. Hudson J C 1969 A location theory for rural settlement. Annals of the Association of American Geographers 59: 365–81
13. Malecki E J 1997 Technology and Economic Development: The Dynamics of Local, Regional, and National Competitiveness. Longman, Essex, UK
14. Meyer D R 1980 A dynamic model of the integration of frontier urban places into the United Sates system of cities. Economic Geography 56: 120–40
15. Mikesell M W 1967 Geographic perspectives in anthropology. Annals of the Association of American Geographers 57: 617–34
16. Mikesell M W 1978 Tradition and innovation in cultural geography. Annals of the Association of American Geographers 68: 1–16
17. Morrill R, Gaile G L, Thrall G I 1988 Spatial Diff Scientific Geography Series, Sage, Newbury Park, CA
18. Myrdal G 1957 Economic Theory and Underdeveloped Regions. Duckworth, London
19. Pedersen P O 1975 Urban-regional Development in South America: A Process of Diffusion and Integration. Mouton, The Hague, The Netherlands
20. Pred A 1980 Urban Growth and the Circulation of Information: The United States System of Cities, 1790–1840. Harvard University Press, Cambridge, MA
21. Riddell J B 1976 Modernization in Sierra Leone. In: Knight C J, Newman J L (eds.) Contemporary Africa: Geography and Change. Prentice-Hall, Englewood Cliffs, NJ, pp. 393–407
22. Rogers E M 1995 Diffusion of Innovations, 4th edn. Free Press, New York
23. Sauer C O 1952 Agricultural Origins and Dispersals. American Geographical Society, New York
24. Taaffe E J, Morrill R L, Gould P R 1963 Transportation expansion in underdeveloped countries: A comparative analysis. Geographical Review 53: 503–29
25. Wagner P L, Mikesell M W (eds.) 1962 Readings in Cultural Geography. University of Chicago Press, Chicago