Key Concepts and Terms

• Classic diffusion theory is associated with Everett M. Rogers' book, Diffusion of Innovation (1962). In its fourth edition, Rogers defines diffusion as "the process by which an innovation is communicated through certain channels over time among the members of a social system" (Rogers, 1995:5). Innovation, in turn, is relative to the adopter, being any "idea, practice, or object that is perceived as new by an individual or other unit of adoption." Rogers found that early adoption was associated with three factors: importance, space, and time. More important places tend to adopt earlier (supported by Walker, 1969; Brown & Cox, 1971; Leichter, 1983). Places closer to an innovation tend to adopt earlier than those further away (supported by Hägerstrand, 1967; Brown & Cox, 1971; and Klingman, 1980). The rate of adoption follows an S-shaped logistic curve with slow increases in adoptions until a tipping point is reached when adoptions accelerate rapidly, then plateau and increase only slowly to reach the last adopters ((Brown & Cox, 1971:551). Types of adopters may thus be characterized in order as innovators, early adopters, early majority, late majority, and laggards.
  • Elements of classic diffusion theory. In Rogers' theory, diffusion requires four elements: an innovation (something perceived as new), a communication system (a transmission system from one individual, group, or society to another), a social system (provides the domain for the diffusion process), and time (from awareness of innovation through to adoption saturation in the social system). Analysis of diffusion must consider each of these four elements. Classical diffusion theory thus rested on the effects of five factors: (1) characterisitics of the innovation itself (ex., complexity, compatibility with existing structure and methods, observability of benefits, relative advantage over existing forms, and trialability - the extent it can be tried out on an experimental or pilot basis); (2) characteristics of communication channels (ex., interpersonal, organizational, mass media); (3) process characteristics (ex., voluntary, mandated); (4) characteristics of the social system (ex., compatibility of the innovation with social norms); and (5) characteristics of the change agents promoting diffusion (ex., technocratic vs. background in organization development; low-level vs. high-level).

  • Stages of diffusion. Classic diffusion theory held that diffusions progress through five stages:
    1. knowledge (awareness of the existence of the innovation)
    2. persuasion (mobilizing positive attitudes toward the innovation)
    3. decision (securing commitment to adopt the innovation)
    4. implementation (operationalization in use)
    5. confirmation (positive outcomes reinforce the process of diffusion)

• Policy adoption theories.Within the field of public administration, diffusion theory was popularized by Jack L. Walker's (1969) study, "The Diffusion of Innovations among the American States." Walker tried to show that the spread of innovations from state to state could not be explained by traditional expenditure models, but instead by diffusion processes.

  • Adoption motives. Berry and Berry (1999) identified three motives leading to state policy adoption: (1) copying what has been a success elsewhere; (2) seeking competitive advantage; and (3) responding to citizen pressure.

  • Adoption models. Policy research following Walker fell into three emphases: (1) determinants models based on political, economic, and demographic factors underlying diffusion; (2) regionalism models based on intergovernmental relationships among states in a region; and (3) federalism models based on the effects of federal incentives and interventions. Using different labeling, Berry and Berry (1999) identified (1) national interaction models focusing on the role of such groups as the National Conference of State Legislatures or the National Association of Governors; (2) regional diffusion models focusing on the influence of neighboring states; and (3) vertical influence models focusing on the influence of the national government.

• Correlates of diffusion have been found to include:

  • Rogers' correlates. Rogers (1995) work found early adopters to be younger, more educated, and have higher SES, have more tolerance for risk, be better networked with others, seek the advice of opinion leaders, work for larger organizations, be less rigid, have more lofty goals, and be more innovation-minded in general.

  • Regional proximity. Early scholarship on diffusion focused on regional proximity as a primary determinant of innovation adoption (McVoy, 1940; Walker, 1969; Walker, 1973). Numerous more recent studies have continued to find a regional diffusion effect (Collier & Messick, 1975, regarding adoption of Social Security by nations; Berry & Berry, 1990, on state adoption of lotteries; Berry & Berry, 1994, on state tax policy; Daley & Garand, 2005 on state environmental policy; and Shipan & Volden, 2005, on local antismoking ordinance adoption).

  • Successful peer adoption. Walker (1969: 897) wrote, "In all cases...the likelihood of a state adopting a new program is higher if other states have already adopted the idea. The likelihood becomes higher still if the innovation has been adopted by a state viewed by key decision-makers as a point of legitimate comparison." Walker argued that when peer states had adopted an innovation, relative deprivation became a motivator for others to follow suit. Gray (1973) the cumulative proprotionof adopters of a given policy to be a predictor of the rate of spread to other potential adopters, based on logistic regression for 12 laws invluding civil rights, welfare, and education. Berry & Berry (1990) also found successful adoption of state lotteries in nearby states increased the likelihood of adoption. Mintrom (1997) found policy innovation in education more likely in a state bordering other states which had adopted or considered a given innovation.

  • Presence of policy entrepreneurs and strong executive leadership. Linking entrepreneurship to innovation is, of course, part of classic economic theory (ex., Schumpeter, 1912). In modern times, Mintrom (1997) focused on the role of policy entrepreneurs in educational innovations. Gray (1994) noted the importance of the role of the governor in adopting economic development innovations. Walker (1969) developed innovation scores for each state and over time, rankings on state innovation have remained very consistent (Gray, 1994). Thus, some units (ex., states) may have a culture of innovation more than others, and this culture may mediate between determinants and adoption (Akers, 2006). Contrary evidence: Some studies fail to find a relationships between entreneurship and innovation adoption. Hoyman & Weinberg (2006) found no relation between human capital and county adoption of prisons to achieve economic development, for instance.

  • Larger organizations have more resources for innovation and innovate more in early stages, but by the time late adopters get on board, the correlation of innovation and size disappears (Mohr, 1969: 126). In general, large organizations tend to be early innovation adopters (Crain, 1955; Rose & Joskow, 1990). Blau & McKinley (1979: 215), however, found that for non-innovative organizations (architectural firms were studied), size was the major determinant of how activities were structured, whereas for organizations characterized by innovation, size was much less important.

  • Larger communities. Mohr (1969) noted that larger communities have larger organizations and so are the locale of more innovation but argued that larger organizations in smaller communities would be equally innovative. Ho (2004), in studying e-government adoption, found that mounting a municipal website was not a function of community size but rather of policy entrepreneurship, but after the relatively resource-cheap effort of establishing a site, the much more resource-demanding matter of populating the site with an extensive range of service features was primarily determined by community size. Contrary evidence: Some studies have linked adoption of innovation to smaller communities, not larger, perhaps reflecting differences in governmental level, policy issue, or dimension (discussed below). Gianakis & McCue (1997) found local government finance officers more likely to adopt administrative reforms. Shipan and Volden (2005) found smaller local governments more likely to adopt anti-smoking ordinances.

  • Socioeconomic factors such as education, income, and urbanization

  • Racial and ethnic diversity

  • Existence of slack government resources. Mohr (1969: 126) argued slack resources free the organization to move from efficiency goals to prestige goals, which are more conducive to innovation. Other studies linking slack resources to innovation adoption include Schumpeter (1912) and Daley & Garand (2005). Contrary evidence: If the innovation is revenue-raising, such as state lotteries, lack of slack resources may motivate innovation (Berry & Berry, 1990). Berry & Berry (1994) also found raising state taxes more likely if peer states had done so, even during fiscal crises. Other studies have simply failed to find a correlation between slack resources and innovation adoption (Brudney & Selden, 1995; Gianakis & McCue, 1997).

  • Existence of national resources. Karch (2006: 418), in a study of human services policy (individual development accounts, family caps, and medical savings accounts), found "state policy enactment is more likely when national intervention provides resources that facilitate innovation." Karch (2006: 403) wrote, "national intervention affects the likelihood that state lawmakers will adopt an innovation by altering the strength of the obstacles that prevent innovation or by providing resources to help overcome these obstacles. National intervention that does not affect these determinants of innovation seems to have no significant effect on state officials’ decisions."

  • Administrative and legislative professionalism. Walker (1969: 898) emphasized the role of professional associations, research centers, and interest groups in establishing norms of best practice, thus sanctioning innovation. See also Balla (2001).

  • Well-developed policy networks. Walker (1969: 897-898) saw information networks as a diffusion prerequisite, fostering the spread of information about innovations and about adoptions by peers. Mintrom and Vergari (1996; 1998: 146) also found "policy networks are important resources that successful policy entrepreneurs draw upon when developing and selling their policy ideas." In their view, policy networks facilitate agenda-setting, which sanctions diffusion of innovation.

  • Political factors have been found to be more important than economic and demographic factors (ex., Martin, 2001; Lindlad, 2006). Mintrom (1997) found policy innovation to be less likely in election years, for instance.

    • Political orientation. As would be expected, innovation regarding a controversial policy is linked to the political orientation of the jurisdiction's chief political officer (ex., governor for states), as Gray (1973) found it was for innovation in civil rights, welfare, and educational policy arenas. Grossback, Nicholson-Crotty, & Peterson (2004) found governors look to ideologically similar states for learning with regard to policy innovation. Likewise, Martin (2001) found political orientation correlated with adoption of lving wage legislation. Contrary evidence: Some studies find no correlation of leadership political orientation with innovation adoption, as Soule & Zylan (1997) found with regard to welfare eligibility reform. Hoyman & Weinberg (2006) found political orientation did not correlate with adoption of prison sitings as an economic development policy innovation.

      Interest group strength. For adoption of state lotteries, presence of fundamentalist religions had been found to reduce adoption (Berry & Berry, 1990), for instance. Interest group strength was also important in accounting for local gun control legislation (Godwin & Schroedel, 2000).

    • Minimizing political costs. Berry & Berry (1994) found adoption of state taxes was more likely when the next election was far off, as legislators sought to minimize political costs.

    • Unified party control has also been linked to policy innovation.

• Dimensions of innovation.Correlates of diffusion may be contingent on the dimensions of the innovation studied. In his conceptual paper, Cooper (1998) faults the correlates of diffusion research for misguided unidimensionality. He argues for classifying innovations on three dimensions and speculates that diffusion may be different depending on where the innovation lies in this three-dimensional attribute space. The three dimensions are (1) incremental versus radical innovation, (2) administrative vs. technological innovation; and (3) process vs. product innovation. Other authors have also noted that the type of innovation affects the diffusion process. Swanson (1994) set forth three types, which further differentiate the incremental/radical dimension later noted by Cooper:
  1. Type I innovations involved changes purely within the realm of a support unit (ex., the IT unit for a computer innovation).
  2. Type II innovations involved changes in multiple supporting administrative units (ex., performance measurement may affect planning, budgeting, and IT units).
  3. Type III innovations involve changes in the core technology of the organization (hence requiring organizational reengineering).

Illustrative Hypotheses

• The larger and/or more important the unit, the more likely it is to be an early adopter.

• The closer a unit is to an adopter, the more likely it is to adopt itself.

• The rate of adoption follows a logistic (S-shaped) curve over time.

• The higher the SES associated with the unit, the more likely it is to be an early adopter.

• Diffusion is faster in urban than in rural settings.

• Diffusion is faster in diverse than in homogenous settings.

• The more risk tolerance associated with the unit, the more likely it is to be an early adopter.

• The better networked the unit, the more likely it is to be an early adopter.
  • The more developed the regionalization for the agency, the faster the diffusion process.

  • The more developed the policy networks in the innovation domain, the faster the diffusion process.

• The more strategic the goals of the unit, the more likely it is to be an early adopter.

• The better the system for communicating innovations, the faster the diffusion process.

• The more supportive of innovation the social system, the faster the diffusion process.
  • The more the innovation is compatible with existing structure and processes, the faster the diffusion process.

  • Adoption is faster for support units than for operational units.

  • The more the innovation requires reengineering of processes, the slower the diffusion.

  • The more the innovation requires restructuring of the organization, the slower the diffusion.

• The more complex the innovation, the slower the diffusion process.

• The more observable the benefits of the innovation, the faster the diffusion process.

• The more the innovation can be piloted rather than only introduced full scale, the faster the diffusion process.

• The more the innovation is mandated from above, the faster the diffusion process.

• The more the change agents associated with promoting the innovation have organization development skills, the faster the diffusion process.

• The more the slack resources, the greater the rate of adoption.

• Diffusion correlates more highly with political factors than with resource factors.

• The more unified the party/political control, the faster the diffusion process.
  • The more top leadership support for the innovation, the faster the diffusion process.

  • The more grants or other incentives provided from above, the faster the diffusion process.

• The more innovation advocates are policy entrepreneurs, the faster the diffusion process.

Frequently Asked Questions

• How is diffusion theory related to institutional theory?
  Though institutional innovations may spread at first for performance reasons, later in the diffusion process they are apt to be adopted for reasons of legitimacy rather than actual performance ( DiMaggio and Powell, 1991). Part of institutional theory, discussed separately, emphasizes the difference between the formal and actual purposes of organizational procedures; formal and actual reasons for adoption of an innovation would be an example of this. Institutional theory also predicts a pattern of isomorphism whereby the early diffusion process is associated with inter-agency differentiation and the later diffusion process is associated with inter-agency homogeneity.

  
  

• How is diffusion theory related to the theory of reasoned action and the technology acceptance model?
  TRA and TAM are variants on diffusion theory, discussed separately. Compared to Rogers' theory of diffusion, TAM places more emphasis on psychological predispositions and social influences on behavioral intention to adopt an innovation. TAM is widely used in business management literature to explain the adoption of computers, Internet use, e-commerce, and other technologies.

  
  

• What statistical procedures are used in analysis of diffusion/adoption?
  Logistic regression has been a common approach, with a dichotomous adoption/not adoption dependent variable, as in the work of Gray (1973). Often maximum likelihood logit and probit have been used with a dichotomous dependent, as in the work of Allison (1984) and Freitas (2006). Berry & Berry (1990) utilized event history analysis. Some have advocated other functional forms, such as complementary log-log (cloglog regression) functions may be more appropriate to diffusion processes than either logit or probit functions (cloglog functions have fatter tails toward 0 and approach 1 more quickly, such that early adoptions are less rare and toward the end, saturation is reached more quickly).

  The standard method has been criticized (ex., Berry & Berry, 1992: 192) for homogenizing the probability of adoption for all years and not taking duration (time to adoption) into account. Modeling duration dependence is seen by some as critical to analysis of diffusion processes. Moreover, logit and probit regression assume observations are independent across time and space. Since this assumption is generally not reasonable, reported standard errors are in error. This problem of logit and probit regression is mitigated if the covariates include a time variable and number of neighbors adopting as a spatial variable.

  When the dependent is time-to-adoption or adoption rate (adoptions per time unit), then Cox regression or some other variant of survival analysis is preferred. Defining the "hazard rate" as the rate of adoption, the traditional logit/probit approach assumes a flat hazard rate, whereas Cox regression and survival analysis utilizes a changing, instantaneous hazard rate specific to the time of adoption.

  
  

Bibliography

• Akers, E. J. (2006). A study of the adoption of digital government technology as public policy in the American states. Dissertation, Auburn University Montgomery, Montgomery, Alabama.
• Allison, Paul D. (1984). Event history analysis: Regression for logitudinal data. Newbury Park, CA: Sage.
• Balla, S. (2001). Interstate professional associations and the diffusion of policy innovations. American Politics Research 29(3), 221-245.
• Berry, F. S. & Berry, W. D. (1990). State lottery adoptions as policy innovations: An event history analysis. American Political Science Review 84(2), 395-415.
• Berry, F. S. & Berry, W. D. (1992). Tax innovation in the states: Capitalizing on political opportunity. American Journal of Political Science 36(3): 715-742.
• Berry, F. S. & Berry, W. D. (1994). The politics of tax increases in the states. American Journal of Political Science 38(3), 855-859.
• Berry, F. S. & Berry, W. D. (1999). Innovation and diffusion models in policy research. Pp. 169-200 in Sabatier, P. A., ed., Theories of the policy process. Boulder, CO: Westview Press. Reprinted in the 2005 edition also.
• Blau, J. R. & McKinley, W. (1979). Ideas, complexity, and innovation. Administrative Science Quarterly 24(2), 200-219.
• Brown, L. A., & Cox, K. (1971). Empirical regularities in the diffusion of innovation. Annals of the Association of American Geographers, 61(3), 551-559.
• Brudney, J. L. & Selden, S. C. (1995(). The adoption of innovation by smaller local governments: The case of computer technology. American Review of Public Administration 25(1), 71-86.
• Collier, D. & Messick, R. E. (1975). Prerequisites versus diffusion: Testing alternative explanations of Socila Security adoption. American Political Science Review 69(4), 1299-1315.
• Cooper, J. R. (1998). A multidimensional approach to the adoption of innovation. Management Decision 36(8), 493-502.
• Crain, R. L. (1966). Fluoridation: The diffusion of an innovation among cities. Social Forces 44(4), 467-476.
• Daley, D. M. & Garand, J. C. (2005). Horizontal diffusion, vertical diffusion, and internal pressure in state environmental policymaking, 198901998. American Politics Research 33(5), 615-644.
• DiMaggio, P. J., & Powell, W. W. (1991). The new institutionalism in organizational analysis. Chicago: University of Chicago.
• Freitas, Isabel Maria Bodas (2006). Exploring differences in the adoption patterns of organisational and managerial innovations for early and late-adopters. Brighton, UK: Science and Technology Policy Research (SPRU), University of Sussex. Retrieved 11/10/06 from http://www.sussex.ac.uk/Units/spru/events/ocs/viewpaper.php?id=52.
• Gianakis, G. A. & McCue, C. P. (1997). Administrative innovation among Ohio local government finance officers. American Review of Public Administration 27(3), 270-286.
• Godwin, M. L. & Schroedel, J. R. (2000). Policy diffusion and strategies for promoting change: Evidence from California local gun control ordinances. Policy Studies Journal 28(4), 760-776.
• Gray, V. (1973). Innovation in the states: A diffusion study. American Political Science Association 67(4), 1174-1185.
• Gray, V. (1994). Competition, emulation, and policy innovation. Pp. 230-248 in Dodd, L. C. & Jillson, C., eds., New perspectives on American politics. Washington, DC: CQ Press.
• Grossback, L. J., Nicholson-Crotty, S. & Peterson, D. A. M. (2004). Ideology and learning in policy diffusion. American Politics Research 32(5), 521-545.
• Hägerstrand, T. (1967). Innovation diffusion as a spatial process. Chicago: University of Chicago Press.
• Ho, A. T-K. (2004). Explaining the adoption of e-government features: A case study of Iowa County Treasurers Offices. American Review of Public Administration 34(2), 164-180.
• Hoyman, M. & Weinberg, M. (2006). The process of policy innovation: Prison sitings in rural North Carolina. Policy Studies Journal 34(1), 95-112.
• Karch, A. (2006). National intervention and the diffusion of policy innovations. American Politics Research 34(4), 403-426.
• Klingman, D. (1980). Temporal and spatial diffusion in the comparative analysis of social change. American Political Science Review, 74(March), 123-137.
• Leichter, H. M. (1983). The patterns and origins of policy diffusion: The case of the commonwealth. Comparative Politics, 15(2), 223-233.
• Lindblad, M. R. (2006). Performance measurement in local economic development. Urban Affairs Quarterly 41(5), 646 - 672.
• Martin, I. (2001). Dawn of the living wage4: The diffusion of a redistributive municipal policy. Urban Affairs Review 36(4), 470-496.
• McVoy, E. C. (1940). Patterns of diffusion in the United States. American Sociological Review 5(2), 219-227.
• Mintrom, M. (1997). Policy entrepreneurs and the diffusion of innovation. American Journal of Political Science 41(3), 738-770.
• Mintrom, M. & Vergari, S. (1998). Policy networks and innovation diffusion: The case of state education reforms. Journal of Politics 60(1), 126-148.
• Mintrom, M. & Vergari, S. (1996). Advocacy coalitions, policy entrepreneurs, and policy change. Policy Studies Journal 24:420-434.
• Mintrom, M. & Vergari, S. (1998). Policy networks and innovation diffusion:> The case of state education reforms. Journal of Politics 60: 126-148.
• Mohr, L. B. (1969). Determinants of innovation in organizations. American Political Science Review 63(1), 111-126
• Rogers, E. M. (1962). Diffusion of innovations First ed. New York: Free Press. Fourth Edition, 1995.
• Rose, N. L. & Joskow, P. L. (1990). The diffusion of new technologies: Evidence from the electric utility industry. Rand Journal of Economics 21(3), 354-373.
• Schumpeter, J. A. (1912). Theory of Economic Development, English Translation of Theorie der wirtschaftlichen Entwicklung NY: Oxford University Press, 1961.
• Shipan, C.. & Volden, C. M. (2005). Diffusion, preemption, and venue shopping: The spread of local antismoking policies. Scholars in Health Policy Research Program Working Papers Series-31. Princeton, NJ: Robert Wood Johnson Foundation.
• Soule, S. A. & Zylan, Y. (1997). Runaway train? The diffusion of state-level reform in ADC/AFDC eligibility requirements, 1950-1967. American Journal of Sociology 103(3), 33-762.
• Swanson, E. B. (1994). Information systems innovation among organizations. Management Science, 40(9), 1069-1092.
• Walker, J. L. (1969). Diffusion of public policy innovation among the American states (computer file). Ann Arbor, MI: Inter-University Consortium for Political and Social Research.
• Walker, J. L. (1969). The diffusion of innovations among the American states American Political Science Review 63(3), 880-899.
• Walker, J. L. (1973). Comment: Problems in research on the diffusion of innovations. American Political Science Review 67(4), 1186-1191.

Copyright 1998, 2006 by G. David Garson.