1. The Diffusion of Innovations
Lecture 15
The Diffusion of Innovations

2. What is Diffusion of Innovation?
“Diffusion is as much a process by which new technologies are developed as it is a process by which usage spreads….”
-Geroski, p.623
However, focus has really been on the spread/usage…definitely not about ‘predicting’ new technologies.
But, a whole heck of a lot about which technology ‘won’ out over some other technology.

3. Definition of Diffusion of Innovation
“the process by which an innovation is communicated through certain channels over time among the members of a social system” (Rogers 1983)

4. Four Basic Concepts in Diffusion of Innovation
Idea, object, or practice that is perceived as new
Channels of Communication
Means by which info is transmitted to or within the social system
Time
Rate at which the innovation is diffused or the relative speed with which it is adopted
Social System
Individuals, organizations, or agencies that are potential adopters of the innovation

5. Common Scope of Diffusion Research
(1) Characteristics of an innovation which may influence its adoption
(2) Decision-making process that occurs when individuals consider adopting a new innovation
(3) The characteristics of individuals that make them likely to adopt the innovation
(4) The consequences for individuals and society of adopting the innovation
(5) Communication channels used in the adoption process (efficiency, speed of distribution, etc)

6. Various Early Diffusion Studies
Investigator
Innovation
Social System
Rapoport (1978)
Radioisotopes
U.S. Hospitals
Perry and Kraemer (1978)
Computer Applications
Local Govt
Pitcher et al. (1978)
Collective violence
Countries
Oster (1982)
Basic oxygen furnace
Steel manufacturers
Adapted from Mahajan and Peterson (1985)

7. Purpose of Diffusion Models
To depict the successive increase in the number of adopters over time.
Permits prediction of the continued development of the diffusion process.
Facilitates a theoretical explanation of the dynamics of the diffusion process.

8. Who uses this stuff? Marketers! Movie studios
Political and Interest groups

9. Foundations of Research on Diffusion of Innovations
Gabriel Tarde (1903)
Proposed the S-shaped curve
As it turned out, study after study tended to show the exact same S-shaped curve when researchers plot the rate of adoption over time.

10. Diffusion “S” Curve

11. Foundations of Research on Diffusion of Innovations
Ryan and Gross (1943)
Categories of Adopters (relative time of adoption)
Innovators
Early adopters
Early/Late Majorities
Laggards

12. Opinion Leadership
Opinion leaders are key for influence and thus ability to successfully diffuse an innovation.
Opinion leaders are concentrated among the early adopters, not the innovators.
In other words, you cannot stack the deck– opinion leaders cannot be ‘created’ as much as they are nurtured

13. Adopter Categories

14. Stages of Adoption Everett Rogers (1995) Awareness Interest Evaluation
Trial
Adoption

15. Categorizing Adopters and Non-Adopters
Adoption:
accept and use innovation
Nonadoption:
nontrial of an innovation
Discontinuance:
rejection of an innovation after it has previously been adopted

16. Dvorak/QWERTY/Beta/VHS/Dos/Mac/grrrrr???

17. Epidemic Models Based on simple examination of “spread”
Simplest version is basic exponential model
Central-source model
N of users
Time
When would we expect this?

18. Epidemic Models (continued)
Modified spread model
Diffusion works through word-of-mouth (i.e., previous users)
N of users
Time

19. Cumulative and Individual Adoption Patterns

20. Assumptions of Simple Epidemic Models
Homophily
Individuals or groups tend to hang out with others who are similar to them (demographics, attitudes, etc)
N is usually constant
Speed of Diffusion usually constant

21. Transmission versus Persuasion
The situation gets complicated when we do not equate transmission with persuasion.
Persuasion may be influenced by several factors– e.g., risk, ‘trustworthiness’ of persuader.
As Rogers points out, this complexity is part of the reason that the S-shaped curve is rarely symmetric

22. Accounting for Adoption Decisions
Probit models
Various characteristics (xi) affects the profitability of adoption a new technology
Not Adopt
Think of X* as shifting to the left– if you do this and plot it, you get S-shaped curve over time.
Adopt X*

23. “Relevant Characteristics”
Probit models depend on specifying relevant characteristics which might influence potential adoption.
Potential Relevant Characteristics (Geroski 2000)
Firm Size as one of the most common– why?
Suppliers
Technological Expectations
Costs
Learning costs
Search Costs
Switching Costs
Opportunity Costs

24. Another Possibility: Information Cascades (Geroski)
What about the innovations that do not successfully diffuse?
“Information Cascades” involve the process of early inertia, potential adopter investment, and the adoption ‘bandwagon’
Three phases:
Initial choice
Lock-in
bandwagon
Photo: engadget.com

25. Rethinking ‘Classic’ Diffusion Models
Taking “the” new technology for granted
S-curves may not just be the starting point of an analysis of diffusion, but rather exist as one possible outcome.