Diffusion of Social Science Innovation into High-Technology Marketing
[Here is a piece that I wrote for the RTWReport that I should have posted before the More on Moore, Pt. I. I summarized Everett Roger's work on the diffusion of innovation that led Moore to adopt the TALC as his major model. I will have to break it into sections. I hope this helps.]
Diffusion of a new artifact--regardless of its obvious merit or whether it is a product, service, design, practice, or idea--remains a risky enterprise. Marketing is a planned process for speeding up diffusion of a high-technology artifact. A first principal is everything takes longer and costs more than anticipated. Successful diffusion of your innovative ideas renders them obsolete. As Dr. William Davidow (1986, p. 145) put it, “Time is the deadly enemy of the technology business.”
The diffusion of artifacts tells the story of human history. Our footprints in time tred from flaked hand axes to the printing press to the Internet. The pace of this march is accelerating. Knowledge begets new knowledge crafted from the Lego blocks of prior innovations. Inherently, high technology designs new artifacts for new markets. New markets require us to use principles derived from sixty years of studying the social processes that diffuses innovation.
In 1962, Dr. Everett Rogers summarized this body of research in his First Edition of Diffusion of Innovation. It became a classic. In Diffusion, Rogers enunciated a set of principles derived from the burgeoning research, mounting from 405 articles in 1962 to over 4000 in 1994 to over 5000 today. According to Rogers (p. xiv), “No other field of behavioral science research represents more effort by more scholars in more disciplines in more nations.”
His Fifth Edition is scheduled for publication in August 2003.
amazon.com
Drawing on Rogers’ Fourth Edition, I want to introduce his framework for understanding the diffusion of innovation, to describe some key experiments, to illustrate some significant ideas, and to close by pointing out a single significant interrelationship with Geoffrey Moore’s high-tech marketing strategy.
Diffusion of Innovation
Rogers (1995, p. 5) proposed that, “Diffusion is the process by which an innovation is communicated through certain channels over time among the members of a social system.” This process of diffusion is characteristically depicted as an S-shaped curve derived from plotting the frequency of cumulative adoption over time. The fours main elements of diffusion are the: (1) innovation; (2) communication channels; (3) time, and (4) social system.
Innovation. Often used synonymously with technology, an innovation is an idea, practice, or object that is perceived as new by a potential adopter. Technology reduces uncertainty in information because it is based on knowledge about cause-and-effect relationships. People may perceive the specific innovation to be part of a technology cluster of closely related technologies, perhaps even organized into an open technology system.
The perception of the innovation influences its rate of adoption. Innovations perceived as advantageous, compatible, simple, and easy to observe or try out are adopted more readily and rapidly.
Communication Channels. Transmitted through a channel, diffusion entails communications about the merits and usability of an innovation. To create awareness-knowledge, mass media channels are rapid and efficient. But, interpersonal communication is at the heart of the diffusion process. Before deciding to try something new, most of us rely on the subjective evaluations of peers who have already adopted it. No marketing campaign can top word-of-mouth. It is the clincher.
Time. The diffusion of innovation is time-dynamic. First, the process of making a decision about an innovation proceeds through temporal stages: (1) Knowledge; (2) Persuasion; (3) Decision; (4) Implementation; and (5) Confirmation. During the persuasion and decision stages of the innovation-decision period, an individual wants to know the innovation’s advantages and disadvantages for his own situation from people who are like him and trying to solve similar problems.
Second, innovativeness is the degree to which an adopter is relatively early or late in adopting a new idea when compared to others in his social system. Rogers’ classified the innovativeness of members of a social system from early to late as: (1) innovators; (2) early adopters; (3) early majority; (4) late majority; and (5) laggards.
Third, the rate of adoption is the relative speed with which members of a social system adopt an innovation. When plotted, the resulting distribution of cumulative frequency over time forms an S-shaped curve.
Social System. A social system is a set of interrelated units, whether of individuals, informal groups, organizations, or segmented subsystems, that are engaged in joint-problem solving to accomplish a common goal. Formally organized social structures often position individuals in hierarchical systems that specify expected patterns of interaction. Informal interpersonal systems are structured by recurrent patterns of self-assortment and characteristic communication flows. Norms develop in social systems that specify acceptable and expected patterns of behavior and interaction.
Individuals may have stakes in influencing an innovation-decision. A change agent attempts to influence innovation-decisions is a manner deemed desirable by a change agency. He has a designated mission; everyone knows it and partially discounts his advice. Within the informal social system, on the one hand, the individual who ranks highest on innovativeness necessarily deviates from group norms, being too far in front to lead. On the other hand, an individual’s technical competence, social accessibility, and conformity to the system’s norms may earn him a position as an opinion leader. Opinion leadership measures how much influence an individual has on the attitudes and behavior of other members.
Classic Experiments
Diffusion of Hybrid Corn in Iowa. Introduced to Iowa in 1928, the innovation of hybrid corn ushered in a cluster of innovations that revolutionized American agriculture between the 1930’s and 1950’s. Hybrid corn was drought-resistant, suited to mechanical harvesting, and yielded about 20 percent more bushels an acre. But a farmer had to change his ways. Instead of selecting his own open-pollinated seeds from last year’s best-looking corn plants, the farmer had to buy new hybrid-seeds each year, because the corn lost its hybrid vigor after a single generation.
With a fresh Harvard PhD in his pocket, Bryce Ryan came to Iowa State wanting to understand social factors in economic decisions. Once there, he chose to study the diffusion of hybrid corn as the current innovation of interest among farmers in two small Iowa communities. After all, much of the innovative research on hybrid-corn had come from the labs in Ames.
In 1941, the Professor hired a new graduate student in rural sociology, Neal Gross, to help him. Neal was a city-boy, but as keen for hard work as any Iowa farmer. When he learned that farmers got up early, he was at his first farm by 6:00 a.m., averaging an incredible 14 interviews a day using a structured survey.
All but 2 of the 259 farmers adopted hybrid-corn between 1928 and 1941. When plotted cumulatively, the rate of adoption over time formed the first S-shaped curve. After five years, only 10% of the farmers had adopted. But, in the next three years, the adoption curve “took off,” with the inflection point in 1935, shooting up to 40% by 1936. The rate then dropped off as fewer and fewer farmers were left to adopt.
When assigned to adopter categories, the innovators, when compared to later adopters, had larger farms, higher-incomes, more education, and were more cosmopolite, having traveled to Des Moine, about 75 miles away, more often. (How do you keep them down on the farm after they have seen the bright lights of Des Moine?)
Although hybrid-corn was a spectacular economic success, the innovation-decision period averaged 9 years from awareness-knowledge to decision-to-adopt. Another 3-4 years passed after planting a small trial plot to confirm the wisdom of the move to 100% hybrid-corn.
The farmers had first heard of hybrid-corn from a salesman, a change agent with his own agenda. But, neighbors were the most frequently cited channels leading to persuasion. According to Rogers (p. 34), “The two rural sociologists intuitively sensed what later diffusion scholars were to gather more evidence to prove: That the heart of the diffusion process consists of interpersonal network exchanges and social modeling between those individuals who have already adopted an innovation and those who are influenced to do so. Diffusion is fundamentally a social process.”
Published in 1943 in Rural Sociology, the Ryan and Gross study became the most widely cited diffusion publication. In itself, it was an innovative research paradigm for studying the process of diffusion, which followed its own S-shaped curve of diffusion, taking off about a decade later in rural sociology and in the 1960’s across several academic disciplines. The cumulative frequency curve of the research paradigm revealed an inflection point of high growth around 1956 that lasted about 20 years before tailing off. More than any other study, this citation classic influenced the methodology, theory, and interpretation of research on diffusion.
PhD-missionaries for diffusion research, like Everett Rogers, left Iowa State for Ohio State and other Universities, and taken together, formed an invisible college. A study of 221 rural sociologists in 1972 revealed a set of highly interconnected links--defined by research collaboration--with two significant cliques of 27 and 32 linked-clusters of researchers. Although unspecified in Rogers, the two hubs were undoubtedly located at Iowa State and Ohio State with Ryan and Rogers, respectively, as their progenitors and opinion leaders. The four largest cliques contained all 8 researchers with 10 or more diffusion publications. The two largest cliques served as network hubs; if they were removed, the network would tend to decompose.
By the 1960s, these investigators had gained enough prominence to begin taking diffusion research abroad into developing countries, to migrate from rural sociology departments into newly forming communication departments, and to diffuse the research paradigm into many academic departments. These network patterns, with their associated power laws for cumulative number of publications by author and topic, are characteristic in academia.
The Classic Columbia University Drug Study. Three sociologists, Elihu Katz, Herbert Menzel, and James Coleman conducted a second classic study of the diffusion of medical drugs. By collecting information on the physician’s social networks, they added a significant addition, sociometrics, to the Iowa State research paradigm. Sociometry measures social relationships, including plotting a sociogram that graphs individual and reciprocal choices. During the 1920’s to 1930’s, first in Europe and later in the US, the innovative Jacob Moreno had introduced Sociometry, along with Sociodrama and Psychodrama. But, it was the Columbia study that established scientifically that diffusion was indeed a social process and that opinion leaders played a key role in the take-off of the S-shaped diffusion curve.
In late 1953, the pharmaceutical firm Pfizer developed the antibiotic drug tetracycline. In early 1954, Pfizer offered a research grant to the Columbia sociologists to find out if their advertisements in medical journey led Physicians to adopt the drug. Setting aside the prosaic question of advertising effects, the sociologist offered more, a rigorous analysis of the social process of diffusion that advanced both science and marketing.
After pre-testing their measures in a pilot study in a sample of 33 physicians in New England, in late 1954, the scientists honed in on a population of physicians in four Illinois cities. The sociologists interviewed 125 physicians, about 85% of physicians in three key medical specialties, “where the new drug was of major potential significance.” These 125 physicians sociometrically nominated 103 additional physicians who were part of their diffusion networks, with the total sample now comprising 65% of all physicians in the four towns. Rather than using recall data from the physicians about time of adoption, the scientists were able to obtain dated records of their first-prescriptions of tetracycline from pharmacies.
Only after the data were collected did the sociologists discover the Ryan and Gross experiment from 15 years earlier. There were striking parallels between the findings from the farmers and doctors. Just as cosmopolite farmers were Innovators so were physicians who attended more out-of-town medical conventions. Just as Iowa farmers classified as Innovators had larger farms and higher incomes, the Innovators among Illinois physicians also had wealthier patients and wealthy medical practices. Hence, Innovators have higher socioeconomic status in both studies.
But, the most noteworthy finding came from the sociometric analysis of the diffusion networks that led to the adoption of “gammanym,” a code-name for tetracycline. Shared professional affiliations within the same hospital, clinic, or practice determined who spoke to whom. The Coleman study (1966) discovered that communication with opinion leaders created a critical mass at the inflection point of the S-curve.
Opinion leaders had received three or more sociometric choices as a “social friend.” During the seventeen month diffusion-period, the opinion leaders had adopted gammanym by the eighth month, the point of inflection. Rogers (p. 68) stated, “…one reason for the S-shaped curve is that once the opinion leaders in a system adopt, they convey their subjective evaluations of the innovation to their many network partners who are thereby influenced to adopt the new idea.”
Rogers continued by indicating that the doctors knew that tetracycline has undergone carefully evaluated clinical trials from drug detail men and medical journals. Rogers (p. 60) contended:
“These communication messages created awareness-knowledge of the innovation among the medical community, but such scientific evaluations of the new drug were not sufficient to persuade the average doctor to adopt. Subjective evaluations of the new drug, based on personal experience of the doctor’s peers, were key to convincing the typical doctor to adopt gammanym for his own patients. When an office partner said to a colleague: ‘Look doctor, I prescribe gammanym for my patients, and it cures them more effectively than other antibiotics,’ that kind of message often had an effect.”
The Nascent Science of Diffusion Networks
In the 1940s and 1950s, a model of mass communication flow, called the Hypodermic Needle Model, postulated that mass media had direct, immediate, and powerful effects on a mass audience. However, a classic study of voters in Erie County, Ohio by Dr. Paul Lazarsfeld and others (1944) during the 1940 Presidential election created a surprising result. Although searching for how the mass media brought about changes in opinions, these scientist discovered that voters’ choices were little influenced by mass media and much influenced by face-to-face contact with other people. From the data, Lazarsfeld (his emphasis) concluded, “that ideas often flow from radio and print to opinion leaders and from these to the less active sections of the population.”
This discovery led him to a Two-Step Flow model, from mass media to opinion leader to others. This still was only part of the story, but it did focus attention upon Lazersfeld’s innovative idea of an opinion leader. What was missing was the idea that different communication channels play significant roles across the various stages of the adoption-process.
Rogers (p. 281) defined opinion leadership as “the degree to which an individual is able informally to influence other individuals’ attitudes or overt behavior in a desired way with relative frequency.” Drs. Paul Lazarsfeld and Robert Merton (1964) used the term, homophily, to describe the degree to which a pair of individuals who communicate are similar in beliefs, educations, social status, and the like. The opposite of homophily is heterophily. Homophily and communication within a dyad or group create increasing returns--each begets more of the other.
Thus, Rogers’ first evidence-based generalization is: “Interpersonal diffusion networks are mostly homophilous.”
As everyone knows homophily can also serve as a barrier to entry from one clique to another. However, heterophilous ties often link two homophilous cliques, spanning two sets of socially dissimilar individuals. Heterophily has special information-spreading potential among diverse sets of homophilous cliques.
For instance, Rao, Rogers, and Singh (1980) studied the diffusion of a new variety of rice in two villages in India, one traditional and one progressive. The opinion leaders in the traditional village were elderly and had little education. Their communication networks were homophilous only within castes, with Brahmins talking to Brahmins, Harijans talking to Harijans, and so forth. Whereas, the opinion leaders in the progressive village were younger, highly educated, and of a higher caste. In the progressive village heterophilous networks links started at the top of the social structure and spread rapidly down across the barrier of castes into homophilous groups.
Rogers offered the following empirical generalizations from six sets of data: When interpersonal diffusion networks are heterophilous, followers seek opinion leaders with: (a) higher socioeconomic status; (b) more formal education; (c) a greater degree of mass media exposure; (d) more cosmopolitan experience; (e) greater change agent contact, and who are (f) more innovative.
What are the characteristics of opinion leaders? Six simple and one more complex empirical generalization emerged from a considerable body of data.
When compared to followers, opinion leaders: (a) had greater exposure to mass media; (b) are more cosmopolite; (c) have greater change agent contact; (d) have greater social participation; (e) have higher socioeconomic status; (f) and are more innovative.
But, opinion leaders are not necessarily Innovators. Sometimes they are; sometimes not. Because opinion leaders who value their role must stay acutely in touch with group norms, Rogers invoked the power of norms to explain this paradox. Rogers (p. 296) noted that if an opinion leader adopts a new idea too quickly, followers might begin to doubt his judgment. One role of the opinion leader is to reduce uncertainty about innovations, but this requires the exercise of pragmatic prudence.
For instance, Dr. Marshall Becker (1970) studied ninety-five directors of local health departments. A low-uncertainty innovation, like inoculation for measles, diffused rapidly, beginning with local innovations by the opinion leaders among the ninety-five health department directors. However, this typical pattern was not present when a high-uncertainty innovation, like a diabetes screening program, was introduced.
The high-uncertainty arises specifically from its violation of physician’s norms. Normatively, diabetes screening was a private fee-based service by local physicians. In this instance, only the directors who were rated as socially marginal introduced this uncertain practice. Less in touch with group norms, they dived into socially risky waters where opinion leaders feared to tred. Only after it was demonstrated that there was no negative fallout from community physicians did the usual opinion leaders introduce the now-made-less-uncertain change, which then rapidly spread throughout the health diffusion network.
Thus, Rogers’s more complex generalization was:
When a social system’s norms favor change, opinion leaders are more innovative, but when the norms do not favor change, opinion leaders are not especially innovative.
From Becker’s study, we can see that, like interpersonal networks, organizations have innovators, opinion leaders, and interorganizational diffusion networks. Innovations diffuse from organization to organization in a process that directly parallels interpersonal diffusion networks.
Dr. Jack Walker (1966) extended this analogy from organizations to the fifty state governments. Each state was scored for its innovativeness, the degree to which a state is relatively early or late to adopt eighty-eight new laws for new state programs for health, educations, welfare, conservation, highways, civil rights, and criminal justice. The adoption of new services required enacting a new law, establishing a new state agency, or offering a new service.
Walker found that New York, Massachusetts, California, New Jersey, and Michigan were the national pioneers in innovativeness. In each region of the United States, however, certain states emerged as opinion leaders, with neighboring states following their lead.
Subsequently, Walker (1971) gathered sociometric data from personal interviews with state officials in a sample of ten states. State officials looked to their immediate neighbors: “State administrators most readily communicate with their counterparts in states that they believe have similar resources, social problems, and administrative styles.” Follower states often copied the exact wording, including typographical errors, of laws originated in a state that served as a regional opinion leader. Iowa is not like New York, but it is like Wisconsin.
Thus, the five states that were national Innovators were too far in front and too normatively different to influence all states. The regional opinion leaders linked the national Innovators to the regional networks, where strong ties within the region led to rapid diffusion. Hence, the fifty states segment their social marketing by region.
If you were to read The Clustered World by Michael Weiss (2000), you would discover that the fifty states could be further segmented, using Claritas’s PRIZM cluster system, into 62 classic lifestyle types, divided into 15 socioeconomic groups, each with demographic descriptions, and neighborhood zip+four codes. Homophilous groups share similar demographics and lifestyles, cluster together in neighborhoods, and to talk to their neighbors.
amazon.com
Each neighborhood is a diffusion network that is linked together by weak ties with their own lifestyle counterparts in the next town or state or around the nation. The world is a Global Village because the clustering principle, common residential patterns within cities, universal human aspirations, the trend toward globalization, six degrees of separation, and the reach and richness created by the Word Wide Web make the world, not a melting pot of “average” earthlings, but a worldwide salad bar that offers diverse choices of lifestyles, but where birds of a feather still cluster together in neighborhoods and affinity chat groups. This core truth remains: you are like your neighbors and those you communicate with most often and most intensely.
Rogers captured this insight in the following generalization:
Individuals tend to be linked to others who are close to them in physical distance and who are relatively homophilous in social characteristics.
A Watershed for Network Links.
The Coleman study was a watershed because it shifted attention away from characteristics of innovative individuals to the study of interpersonal connections in diffusion networks. In his study of the diffusion of tetracycline, Coleman identified seven network links: hospital affiliation, frequent attendance of staff meetings, sharing office space, sociometric choices as a source of information and advice, sociometric choices to discuss cases, sociometric choices as best friend, and reciprocating sociometric choices by other doctors.
Highly connected doctors adopted tetracycline rapidly; social isolates took much longer. The degree of network interconnectedness was a better predictor of innovativeness than any of the individually focused variables. Among the seven, the best predictor proved to be the friendship variable. Coleman’s explanation invoked a social snowball effect that cascades from an innovative friendly doctor to, say, two other doctor-friends, who, in turn, influenced four others, and the like. Rogers’ generalization stated:
The network interconnectedness of an individual in a social system is positively related to the individual’s innovativeness. |
