II.C. Reality Construction as a Product of Delphi Interaction
D. SAM SCHEELE
The mind is but a barren soil;
Reality is a name we give our collections of tacit assumptions about what is. We bring along these realities to give meaning to our interactions. Each of us maintains several of these realities--at least one for every significant set of others in our lives. We have domestic realities, parental-family realities, professional realities, sexual realities, organizational realities, stylistic realities.... Since this article is about Delphi design, the important thing is not how many different realities we each have, but that one important product of each Delphi panel is the reality that is defined through its interaction.
Realities can be described as presumed agreements which give meaning to our thoughts and make reasonable our actions in each setting. Most of these agreements about reality are implicit, and are merely confirmed and elaborated by our acts and conversations. Sometimes our interactions subtly modify these realities. Occasionally, a group's reality is actively renegotiated or even constructed de novo for a new situation. Delphi inquiries might produce any of these results. The purpose of this essay is to suggest ways of managing Delphi interactions in order to create intentionally a reality that will prompt the appropriate kinds of active interventions.
I believe with others that there is nothing more practical than good theory. Much of what is related in this article is theory, but theory in search of application, by the reader, out of his understanding, to produce results in his specific contexts. I have used a number of examples to illustrate how realities are asserted, modified, reconceptualized during a Delphi interaction. Each example will require detailed consideration. This is painful if the reader is interested merely in an overview. Therefore, I have tried to write the discursive text so that it makes sense even if the reader skip the examples. Each example is a freeze-dried caricature of a set of rich interactions. To reconstitute, the reader must supply the cerebral juices and attention, such connective interpretation being necessary to make static diagrams into what might pass for interaction.
Another problem with examples is that they represent only a small fraction of the myriad of potential applications for this approach to Delphi inquiry. Clearly it would be incestuous to build a design rationale solely on generalizations drawn from available applications. Taking an expository approach based on cases would foster an already widespread predilection-method and technique in search of applications. This would be the antithesis of my primary recommendation: that the particular qualities of the circumstances that prompt and define the inquiry be used as a basis for the Delphi design. Further, let me suggest that the results of a Delphi be seen as the product of a carefully designed and managed interaction and not answers to a set of abstract questions that are obtained by following prescribed methods. Hence, a slogan for this essay: Concepts from doing.
This paper might have been called: What to think about when considering, designing, and managing (even interpreting) a Delphi. The reader will find many propositions asserted that require a reflection and reinterpretation for application to his particular undertaking. The extensive illustrations are intended to enable the reader to develop a feel for the importance of details of style and tone in presenting materials to panelists. These illustrations should not be thought of as "cases" to emulate, but as necessary to describe the more general pedagogic points about the importance of self-conscious presentation of information in suggestive, but open-ended, frameworks to facilitate the negotiation of realities. Most of the illustrations are based on Delphis we have conducted. In several cases the substantive content of the illustration has been changed because of the proprietary nature of the inquiry or the possibility that our intent would be misinterpreted if the material were to be seen out o€ context. Also included is some of our thinking which occurred when we did not do a Delphi.The italic text provides a setting for some of the illustrative examples. Each illustration depicts a synthesis of the interaction between the panelists with summarization, juxtaposition, interpretation, reconjecture by the Delphi monitor. The role of the diagrammatic presentation of the examples is described in the illustration below (Fig. 1). The intention is not to create order or to impose a unique conceptualization. Neither are the diagrams supposed to be balanced, "well-designed," synthesizing abstractions, or even documentations. In some of the Delphis, the major part of the panelists' comments were sent back on tape cassettes. The emphasis is a personal verisimilitude with the process of undertaking conceptual forays. Most of the panelists' thinking processes cannot be directly shared, so we have attempted to depict for the group some typical points of view out of which to define a reality of relevance.
The particular graphic style adopted in this paper is a personal one. It developed out of use. It is intended to support the process of thinking and not simply to represent completed conceptualizations. Graphic aids can be useful in stimulating your own thinking and organization of ideas. Start by trying your hand at whatever seems cogent to you and make adjustments as you go. There arc a rationale and some techniques that have evolved in the use of graphics in thinking, but their explanation would make another book. The quality of the drawings is intended to convey the liveliness of the concepts and their important properties effectively and quickly. This is difficult to do in writing or with more formal diagrams. Also this drawing style encourages participation, and the organization of the diagrams usually readily admits of modification or extension. In addition to aiding individual thinking, common graphic constructions, or explicit group memories, are useful in moderating discussions. Here the shorthand of positional relationships and the insightfulness of successive interpretations and alterations prove very productive. Since Delphi inquiries are group processes, these kinds of graphic representations have proved equally valuable in Delphi applications. The figures may not mean the same thing to you as to me, but your explanation is accessible, and therefore should be more useful to you than mine would be.
Some of the points may seem trivial-like, "use bright colors" or "state ideas in emotive language"-but their impact is significant. Others-like "provide a concrete situational context" or "depict an explicit theoretical framework"have resulted from trying to de-abstract Delphi inquiries. It is important to deal with the different assumptions of panelists, monitor,1 and sponsor which, when left alone, limit the potential fruitfulness of the Delphi interaction. Most of the methodological insights suggested here have resulted from efforts at designing other kinds of subjective information processes, such as diffusion of innovations, learning by doing, technology transfer, policy development, management of creativity, and design of service systems. Look before you leap, but eventually leap.
Concepts of Reality
The impact of one conceptualization of a situation upon others and the influence of the various constructions of reality assumed by the Delphi panelists generate what I believe are the most significant results from any Delphi inquiry. Panelists can be made aware of these seemingly subtle differences in the nature of the realities they presume in the course of the interactions. The panel can then produce a common reality for the situation at hand as a result of their participation.2 How this comes about is determined by the monitor of the Delphi interaction. If panelists are reluctant to make specific contributions or if a very wide, almost unrelatable, array of conjectures is produced, one suspects that there are great differences in the meaning each panelist is attributing to the way the inquiries are stated in the materials provided to stimulate response and the way panelists expect the results to be used. This ambiguity often might be what you want-productive of interesting premises. An array of differently bracketed realities that include a particular object event-concept is often useful. For example:
Four young adults who are retarded enter a restaurant with older couple. Panelists were asked to select likely responses for restaurant manager, waiters, and other patrons from a list provided. Many panelists added their own. The panel included parents of, and professionals who work with the retarded, as well as individuals to simulate response of general community. The responses of the panelists could be mapped:
Later, panelists were asked how repeated contact with the retarded would affect the key actors:
On the other hand, you may want to create greater focus and consensus. If this is so you can begin with examples for interpretation instead of general questions. This enables you to direct attention in subsequent rounds to contrasts between the assumptions imbedded in the initial situations and panelists' contributions. Differing reality constructs can produce divergence from even seemingly unambiguous statements. Focusing attention on differences in the reality constructs will usually yield either a more refined and widely agreed upon definition of the appropriate context or clearer and more precise distinctions between competing contexts-possibly leading to an estimation of the relative probabilities of each, or a search for present options that could influence the circumstances.
In the preceding discussion the notion of socially constructed or intentionally negotiated realities was employed. This concept grew out of the work of Husserl,3 Merleau-Ponty,4 Heidegger,5 and others, which led to the formulation of a phenomenological epistemology that is now being applied by neo-symbolic interactionalists and ethnomethodologists. The concept of a negotiated reality can be related to Mitroff and Turoff's discussion of the philosophical bases for inquiry systems in the preceding article. This discussion describes a range of inquiry systems (IS) using as differentiating labels the name of the principal philosopher whose concepts undergird each approach. The array includes the Leibnitzian, Lockean, Kantian, Hegelian, and Singerian IS's. Since these categories are well defined there, the philosophical premise for an IS based on a view of reality as a context-specific product of interaction will be described in relation to this framework. First, to select a label consistent with the others will be slightly misleading, because any one name tends to obscure the contributions of others and imply that the ideas are largely set. Dubbing this territory of philosophic exploration after Merleau-Ponty seemed the least misleading. To make a contrast with the Singerian analyst, the Merleau-Pontyean is concerned with the particular reality created by the "bracketing" of an event or idea out of the great din of experience, rather than explicating a pragmatic reality that can be used to define possible actions. Truth to the Merleau-Pontyean is agreement that enables action by confirming or altering "what is normal" or to be expected. By contrast, the Singerian views truth as an external articulation of systems to define goals and options for action. Reality is viewed by the Merleau-Pontyean as the product created out of intentions and actions instead of an external basis for intelligent actions. To reiterate Mitroff and Turoff, the importance is not which philosophy is "correct," but which is appropriate to the kinds of situations one is attempting to impact. The Merleau-Pontyean inquiry system seems applicable to situations either where a redefinition of contextual reality can facilitate the generation of new options, or where the acceptance of a new reality must be negotiated to create the impetus for technical or social change-as, for example, in defining as "progress" a reduced or more limited consumption that would permit reallocations instead of "progress" as lower unit production cost to support increased demand. This philosophical point of view leads to viewing the future as a situation where both the dominant reality and the technology are invented as well as inherited, and where culture is transformed as well as transmitted.
Merleau-Ponty and others suggest reality be viewed in a new way: as a currently prevailing shared assumption about a specific situation. This implies that reality is the product of our experience and not external to it. In a commonsense view, reality is a collection of observable things and occurrences which is animated by a society of individuals. Although we are not usually aware of these distinctions, our everyday realities can be seen as created by us out of the meanings we give things and events.6Since we do not exist alone, we are continuously asserting and having validated or challenged our definition of "what's going on" or "what it's all about." Our collection of situational definitions constitutes our reality. We select realities from our repertoire that seem appropriate in order to know how to act, attribute meaning, and interpret behavior. This means that instead of continuously discovering more of an external verity-"the reality out there"-we are, wittingly or not, continuously adding, verifying, or revising our "shelf-stock" or versions of what is normal or to be expected in particular circumstances. We each have a shelf-stock of realities that have been produced by our interactions.
Earlier the basic philosophic question had been "What is the structure of social reality?" Now phenomenological insights have transformed this question into "What realities have been or are being socially constructed?"
What does this mean for conducting Delphis? First, since the results of a Delphi are produced by interaction, albeit highly structured, the results can be said to constitute a reality construct for the group.7 Because processes of successive refinement, like the Delphi, strongly tend to induce convergence and agreement, the monitor of a Delphi should purposely introduce ambiguities, even disruptions. These might take the form of "angle" items8 to challenge and redefine reality as well as "quirk" items9to act as catalysts to explore the limits of the reality. For example:
Mass transit could compete with private vehicles by offering more than lower cost particularly in enhancing the use of commuting time by offering:
Round 1: What types of services?
Round 2: Can't money change hands-transit as market place? (angle)Relate services to attracting and serving youth? (quirk)
Round 3: How might the service be organized and supplied?
Second, since the knowable reality is in competition with the other conceptions, including the idea of reality as a negotiable construct, the unknown or unexplained cannot simply be attributed to greater degrees of complexity (i.e., the "more data and better instruments" gambit). This means that further efforts to obtain information, such as a Delphi, must go beyond attempts to unravel what has often been assumed as merely additional complexities. Instead, information should be sought that can shape reality, such as identifying new considerations or introducing new options. This means that the systems being described are viewed as indeterminate, arbitrary, delimited, multiplistic, even convenient fictions if this facilitates discourse, but not as complex Cartesian clockworks. In conducting a Delphi then, "what if" and "why not" items might be introduced or highlighted if suggested by a panelist to prompt consideration of new alternatives.
Third, the reality we construct can be expected to be different by at least as much in the future as our technology will be advanced or our society restructured. This is almost always overlooked by forecasters and other futurists. Predictions may well occur as forecast, but their occurrence will not necessarily mean the same thing then as it now seems that it would. You can note people's naive understanding of this in their response to the prediction of new occurrences with the statement, "...but I guess they [the people of the future will be ready for it by then."
Fourth, expect reality to continue to be negotiated. This means that the kinds of realities within which occurrences will be given meaning and be understood will vary from those prevailing at present. To a large extent changes in reality shape the kind of attention, consideration, and effort that will be spent on developing a new idea. They also determine whether the new concept seems plausible, desirable, and feasible. Further, both interest in, and advocacy of, a new concept, along with precipitous events that come to be associated with it, can shape reality to the extent that a new concept becomes one of the ways that reality is defined. Our present notion of the "urban crisis" is an example of a concept that has become imbedded in our realities.
A Delphi inquiry, then, might explore two sides of the negotiation of reality with regard to a specific occurrence: (1) how alternative realities might affect the meaning of the occurrence and how likely each is and (2) how public perceptions of the occurrence, the interests and activities of its proponents, and the concepts and ideology that come to be associated with it will shape the reality that is negotiated.
In many cases it also may be useful to consider the possibility of precipitating events. These events (seen from hindsight) have often dramatically altered the collective awareness of consciousness of the society. For example, contextspecific realities were shaped by the assassinations of the Kennedys and King for gun-control measures, Nader and "Unsafe at Any Speed" for auto safety campaigns, and the Soviet Sputnik for the U.S. space program and educational redirection.
Let us look at how these considerations can be handled as part of a Delphi inquiry. To grasp how different prevailing realities might affect a particular topic, one can posit several alternative "reality gestalts." Encourage participants to supply their own realities. But 1 believe examples are necessary to clarify this notion initially--even if these proto-realities are overly "caricatured." The example below illustrates this kind of inquiry.
Also, you might want to probe the participants' insights regarding items that might be significant in the construction of realities in the future. Here it is useful to suggest items that you believe are important as possible examples. Even include candidate-precipitating events-although their nature and timing are by definition unanticipatable. Knowing that the meaning of events is, so to say, up for grabs can sensitize managers to the use of public-attitude-shaping tools. Introducing these perturbations will begin the discussion and evoke additions and comments from the panelists. An important corollary point is: the meaning of the future in the reality of those taking actions (of necessity in the present) may change. It has dramatically in the past ten years or this book wouldn't have been considered for publication, possibly not even written. One's vision or concept of the future has not always shaped reality in the same way. It is as important to know in what ways images of the future are given meaning as to have a full complement of alternative futures (see Chapter VI, Section D).
The medieval glory-beyond-the-travail-down-here view of the future would prompt a different kind of action from the expectation of a better tomorrow through hard work and technology which has characterized the industrial revolution.
Mental Climates and Styles of Thinking
The dominant reality of the "modern" world has been called the positivistfunctionalist view. In this view reality is "out there" (we are in it, but it is not in us), discoverable by improved access and greater attention and presumed to be ultimately knowable-given sufficient time and diligence. Lukacs10 and Panofsky11 call these global realities the habitus mentalis of an era. A habilus mentalis has at least these three characteristics:
At a macro scale the dominant habitus mentalis of society is being negotiated. Ours is now. I would like to summarize some of the changes we expect in the postindustrial society. To begin, I'd like to share a name we have coined for the postindustrial society that is descriptive of it, instead of merely indicating after industrial. This new label is "the idiomergent era." This is a neologism. Idio is a Greek root indicating uniqueness, separateness, distinctness, as in the word idiosyncrasy, and in its most peculiar sense, idiot. It has been combined with the Latin root mergere, to cause to be swallowed up, to immerse, to combine or coalesce, to lose identity. The result describes a more separate and a unique kind of interdependent aggregation. It seems that the idiomergent era will be characterized by both extreme variety and increased uniqueness in individual behavior and organizational purposes and at the same time be dependent on the greater integration of the individual and organization into the structures of a variety of specialized communities of interest, producing networks of cohesiveness for the society as a whole. Thus the term "idio-mergent" denotes greater individuality and autonomy combined with a deeper involvement of the individual in fluid social arrangements, innovative organizational structures, and experimental personal relationships. This describes change in the vanguard of society: a new habitus mentalis that will spawn new realities.
The habitus mentalis for the idiomergent era will no doubt affect the design of Delphi inquiries. Exactly how depends on the response of practitioners and monitors to their perceptions of what is relevant and their selection and advocacy of ideas that come to be in good currency. I will try to share my construct for this new mental climate. You realize that I have no way of knowing the future, nor am I merely asserting an explanation to serve as a basis for judgment. These and many other possible descriptions of what this image of the future is can be offered. My view is that I am engaged with you in negotiating what to expect from Delphis. Use it as you will.
Before getting to differences in the epistemology, social theory, and guidance process of the thinking style of the idiomergent era and what these augur for the design of Delphi inquiries, I would like to suggest another feature of this era -an increasing concern with the future. Greater interest in the future has resulted from a series of changes in our dominant social reality. "These changes seem relevant to consider in designing systems for assembling useful information. First, change became accepted as normal. Reality became dynamic, but this alone could not make the future more important now; things would just work out differently. Second, technological development and social and physical mobility have increased the number of options available to decisionmakers. This made their reality wider, but not necessarily extended deeper into the future. Third, the number of self-evident paths, inevitable selections, and unquestioned preferences was reduced by an illusion of choice fostered by great increases in the availability of education-both schooling and media-supported awareness. This ballooning of information has expanded reality to include the possible, the imagined, the dreamed, the fantasied, and all other "residents of the future in good standing."
The itch for a fix on the future afflicts both people and organizations. Individuals want to know more about the future so they can choose places to live, things to experience, roles to adopt that will be highly desirable when the future occurs. Businesses analyze .future opportunity areas and establish organizations to prepare to exploit them. Institutions assess their roles in the future and begin reconceptualizations. Government investigates the future and identifies many unacceptable contingencies that require more controls. In most cases the future reality is portrayed as an unfolding panorama of large-scale changes in general circumstances and situations whose meanings are anchored in the present. The inference to be made for the sponsor of an inquiry into the future was, "If I know more about what's going to happen, I can get ready for it." It was like spying on or getting the drop on the future. "These futures were not the world of will, action, and events; but broad backdrops of scattered possibilities from which a series of succeeding "presents" would coalesce on the moving pointer of time. The message was that to "realistically" take effective actions the future must be considered. This future dimension of reality, now so important, has been chiefly portrayed as a passive set of probable possibilities.
The growing recognition that the "meaning" of things and events is not determined, but in fact refers to social agreements that have changed in the past, and which can be renegotiated, even for each circumstance, marks the beginning of the idiomergent era. The reality of the future as a broad techno-cultural script is too ambiguous to support particular undertakings. Attempts have and will be made to bracket specialized future realities that contain only highly relevant possibilities-but in most instances the number of considerations in even a "bracketed" future will not support finite analysis to select a single course of action. But the process can be reversed. Decisionmakers can attempt to negotiate a specific reality for the future which, if they are successful, can come to be widely enough shared that it is realized through successive choices. The future will be invented. Thus in the idiomergent era you can expect attention to focus on the selection of methods to create and portray new realities that have the potential to be actualized through tacit agreement between the intentions of those affected.
In order not to lend credence to the prevailing presumption that Delphi inquiries are concerned only with the future, let us return to look at ways that Delphis can be used to contribute to the store of knowledge, enlarge the society's understanding of itself, and improve the style of governance in an idiomergent culture. For each of these three characteristics of an era, the currently prevailing concepts will be contrasted with their idiomergent counterparts. This will be done rather telegraphically to provide a background for an example of an item in a Delphi inquiry. Elucidating these points more clearly would require another book.
Differences in Epistemology concern not only what is sought and accepted as knowledge, but how information is categorized and organized to support actions. As a man thinks, so is he.
To illustrate these epistemological differences, the best example seems to be the development of categorizations of consumers for use in market planning for new products. In the industrial era, consumers were defined demographically, geographically, and by aggregates of features that indicated a consumer's relative propensity to buy either specific types of items or to buy in a particular way. One idiomergent approach would be to develop first a model for the process of product introduction, next those categories of consumers likely to be involved, and then where and how to locate them and their probable numbers.
Differences in Social Theory are alternative views of how the society is organized, its tenants perpetuated, and its actions explained. Society is a name for us to call "what is."
For the Delphi inquirer these differences in social theory suggest greater emphasis on finding out about the appropriateness of societal norms, roles, and institutions, limning how they came into being and, probably most importantly, suggest ways they might be reshaped. The example describes an application to one particular field.
Round 1: Please elaborate on the diagram below-additional relationships between the retarded and blind person living together to provide mutual support.
Round 2: The relationships described by the panel have been aggregated below; suggest where indicated which significant others might contribute to make this relationship more workable, and what they might do. As a neighbor living in the same apartment building, what could you do to contribute to this relationship?
Round 3: Identify potential dtffculties and problems that might be expected to be encountered.
Round 4: What might you propose to avoid, ameloriale, or resolve these difficulties?
Differences in Guidance Processes used in a society reflect implicit, usually unwitting, agreements about the mutual concerns of groups (from pairs to organizations to segments of the society as a whole). These concerns and their mutuality change. Sometimes they change in response to awareness of threats and opportunities. At other times, improvements in techniques such as communications, media, information systems, or analytic schemes change the way fundamental relationships are perceived. Economically and politically the negotiation of power continues. The society's guidance process is not simply decreed--as by a constitution or framework of laws. It includes these, but it is built up rather than laid on. A society's guidance process embodies in conventions all of the expectations of individuals concerning how things ought to be "fairly" and "knowledgeably" decided in order to produce the greater good and what to do if it's not happening as it ought. Societal guidance is all we ourselves are not contending about and in which we assume someone else is watching out for our interests.
The changes in style of societal guidance that I expect to characterize the emergence of the idiomergent era will contribute directly to the need for more Delphi and Delphi-like processes for collective consideration of topics. However, evaluation of the usefulness of Delphi inquiries will be made more on their suitability as a process for discussion and choice than on the concerns about the accuracy, insightfulness, or agreeableness of the results. The next example was chosen to illustrate how a Delphi can be used to focus consideration of a topic that has no established institutional advocates.
Round 1: What actions are possible to reduce peak impact demand for transportation to reduce "overhead" costs of quality urban lifestyle?
Round 2: Select most promising approach(es)
Round 3: What interests and organizations would be involved in implementation and what interests do they have which could be appealed to?
Round 4: Suggest administrative mechanism to manage program to achieve objectives.
Round 5: Consider implementation: What kind of demonstration program or other approach might be used to rally and mobilize support?
Group Influence on Reality Construction
We all have highly idiosyncratic experiences, ideas, and fantasies. While they actually are our own, their meaning is created in the crucible of their interaction with what is going on in our various contexts. For example, a drearr shared with people you have recently met has a different impact on the reality being built than the same dream would have if described to an informal caucus during a psychiatry conference. In the first circumstance, your reality is being created from such elements as your image as a person, the meaning of a relationship with you, the place of personal interpretation in your respective motivational structures, and the bearing of dreams on events to come. In the second circumstance, the dream becomes material for negotiation of professional prowess, theoretical differences, clinical inferences, concepts of consciousness--even a trigger for discussions of what this discussion group or the profession is all about. You can imagine that sharing facts, feelings, and proposed actions would also impact differently on the reality being constructed in these contrasting contexts. Try imagining the contrasting reality that could be created in each of the two contexts just described (telling potential friends about yourself vs. a discussion about you as a topic in a professional group) for each of the following: the recent death of a relative; your interest in traveling to India; an idea for creating a "kitty" for vacationing by selling the work of local artists to your fellow employees.
Delphi inquiries are conducted with groups where the individual participants might be expected to view the group differently, particularly since they are often anonymous. For most, "being on a Delphi panel" has no particular meaning independent of the topic. Frequently the presentation of the inquiry items and other information is indefinite or ambiguous about the nature of the panel as a group. This directs the panel's attention and energy, in part, to the task of defining the reality of their relationship instead of creating richer realities for the domain of the proported topic. Further, the quality of the reality within which the individual inquiry items are elaborated varies with the ways the panelists view the meaning of the group and its findings, insights, recommendations.
This means that it is not enough that the panelists share a firm idea of the group's identity, but the perception of that identity tends to shape the nature of the individual messages, the quality of the interaction, and the character of what is produced. Table 1 (below) presents a crude taxonomy of the features of groups and some notion of how a group's conception of itself affects the products of Delphi interactions.
The monitor (or modulator) of a Delphi inquiry strongly shapes which conceptualization of the group is assumed by each participant. This usually implicit concept of the group dictates the mode of interaction that each finds appropriate and determines the reality each uses as a frame of reference in attempting to contribute. Several different assumptions can often be made depending on how individuals interpret the implications for reality of the messages in the communications and materials provided them to begin the interaction. Until each panelist is comfortable with his notion of the group's nature and believes it to be confirmed by responses, it is difficult to produce meaningful contributions. In the paragraphs below, I suggest some ways in which a Delphi monitor can shape the group's conceptualization of itself and adopt a mode of interaction that can produce results which will be congruent with their anticipated use.
Transactions will likely become the dominant mode of interaction if there is:
In adopting this mode panelists assume that the participants have no other commonalities and expect none. For most this means they will adopt the most familiar model of interaction-probably "answering a questionnaire as 'accurately' as possible." Exchanges will tend to be formal. They will center on "proof" or "refutation" of ideas identified by the monitor. There will usually be a statistical integration of the group's assessments. Anonymity will likely be used to reinforce the panelists' self-
concept as impartial observers in an "analytic" reality. Panelists who are uncomfortable with this reality will devote great attention to selected issues, often digressions from the principal considerations.
Variations in the transactional mode can be effected by (1) beginning the Delphi with open-ended, agendaless items in search of meaning (to explore the agendas and gestalts of the future held by the panelists); (2) selecting the extraneous and derivative responses of the panelists for feedback and later exploring the interpretation of the discourse with the panelists (to identify latent options, issues, and considerations); (3) extracting the explanatory premises from the panelists' responses after two rounds, then focus on estimation, followed by attempts to elaborate a system of relationships (to build theories and models from premises derived from content rather than technique); and (4) starting with a preliminary design of a thing or program, proceed to iterate critique and redesign to produce a better design by successive interaction of multiple viewpoints, or to derive by interpretation of the "designer-panelists"' responses which considerations seemed important or even critical to the design.
Obviously, there are many other manipulations that could be developed, but these are sufficient to illustrate the type of variations that can be created for particular applications. A Delphi should not be undertaken to validate concepts which you already have developed and refined; panelists want to make significant contributions and these will seldom build meaningfully on highly elaborated initial concepts.12 Delphi inquiries are obviously applicable to much more than obtaining pooled judgements about particular options.
Experiences become the prevailing mode of interaction when:
Inherent in the assumption of this mode of interaction is a commitment to the anticipated process and results. The principal interest of the panelists is the particular qualities or insights each seeks to contribute. Even when anonymity of panelists is maintained, their messages tend to be informal and approval- seeking (other-directed conjectures) instead of raising formal distinctions and striving for accuracy and defensibility. Panelists in later rounds will attempt to make contributions that will affirm their brilliance or uniqueness. The result can be omission of important but obvious points and dangling insights (the monitor can point out these). When the panel is largely drawn from a single discipline or field of application, the interaction quickly moves out to the ethereal zone instead of enriching the context for action. Modifications by the experiential mode include: (1) try to organize the panel into "teams" to represent particular interest groups or points of view (to preview negotiations or develop highly detailed but still preliminary agreements for consideration); (2) present initially a clear and detailed description of the expected product of the inquiry; or a unique and special situation that calls out for great imagination and close collaboration to be described first so that the prevailing rules for judging the suitableness of contributions are partially suspended (to spark creativity and develop a relatively complete conceptualization of a viable new approach).
Episodes are the characteristic mode of interaction for groups that:
This mode of interaction has the highest emotional content and the most potential for prompting action based on the insights produced during the Delphi inquiry, which is by definition supplemental to the group's normal interactions. In these cases you should consider face-to-face group processes. Other techniques (such as program-planning method-PPM, nominal grouping, multiattribute utility assessment) may be as productive, require less time, and have more spillover benefits than a Delphi. It is also imaginable that a group of strangers could sufficiently internalize instructions and materials that would be dramatic enough to induce an episodic style of interaction for an inquiry. Such an inquiry would have to be virtually a simulation. The substantive insights produced by a small purposive group can be expected to be highly specific to their context. The insights produced by such groups regarding the personal and emotional dimensions of the topics are probably not generally indicative. In fact, it may be necessary to create a simulated purposive group in order to probe the interpersonal and emotional dimensions of circumstances that are themselves the subject of conjectures or competing proposals.
Variations on this mode of Delphi inquiry are limited because alteration in the premises usually induces the group to adopt a new reality resulting in a less highly bonded mode of interaction. The most interesting results are produced in instances when the group's strongly shared reality is collectively modified in response to new insights and even to incorporation of new members. However, this collective modification of reality is not always completed successfully and it can be expected several realities will prevail. This can result in dissolution of the group, a breaking down of the group process, or the production of highly emotional and idiosyncratic contributions.
Events, often with more form than content, distinguish the interaction of groups which:
The inherent one-dimensionality of the interactions of these groups exhibits itself most clearly in the importance of agenda-setting. Whether highly informal-"wait, should we be discussing this"-or employing a very formally maintained docket, these groups find it essential to screen topics for relevance to the organizing commonality. Issue focus has become more necessary as society becomes more diverse, in order to keep an affiliative group from fractionating into several puposive-type groups, each of which would maintain multidimensional commonalities. To survive and grow, however, such affiliative groups usually have to deal with threats and opportunities that are on the perimeter (if not outside) the group's reason for being. 'This means that most of the messages exchanged can be assigned to one of two distinct classes: proforma statements necessary to group maintenance and expository statements intended to modify the group's prevailing notion of their common interest. There will be very few exchanges about things that affect them as individuals or alter the group's relationship to the society.
Delphi panels who think of themselves as affiliative groups tend to proceed in two-step responses to the items in the initial inquiry. First, they make narrow and direct responses to each item. Second, they have additional considerations. The whole panel is usually asked whether these should be dealt with. If the panel or some substantial subgroup agrees that the additional items should be added, then they are taken up. Consequently, much of the reality that is created has to do with the meaning of considering certain topics and not the meanings derived from actual consideration of those topics. A Delphi should never be attempted with a "single-purpose" group until they have agreed that there is something that needs their attention which would not take care of itself.
Affairsare the interactions of groups characterized by:
The beginnings of these groups are abstract categorizations which have meaning for the inquirer or the explainer, but seldom for the group itself. So when asked almost anything, these groups scramble in several directions in search of cohesive interests. The direction of a question or inquiry presumes a meaningful response. A variation of this occurs when the inquiry is directed to explore differences within the group. In this instance the inquiry presumes distinct factions. The group will often develop these "factions" in formulating responses, but the factions are transitory. A corollary dynamic in the interactions of these groups is the prefacing of comments with remarks which either attempt to identify the speaker with the whole group or a major faction, or will uniquely distinguish him and his contentions. These interactions tend to be characterized by a series of declarations-"Run it up the flagpole and see who salutes." Occasionally in short interactions, and eventually in almost all, a series of declarations will coalesce a faction (occasionally including the whole group), which will concentrate thereafter on the construction of a position and purpose. Once constituted, such factions interact as large affiliative groups, usually in a way that produces schisms and dissolution.
Because researchers, administrators, and most of the rest of us have been almost conditioned to think of society as divided into distinct kinds of "labeled" groups, the response of abstractly defined groups (e.g., doctors, women, radicals, educated) is frequently sought in inquiries which end in producing results that do not satisfy the inquirer. One way to remedy this is not to expect a coherent position to emerge, but to look for a pattern of diversity in the group's assertions and the ways in which these points are qualified. These groups cannot examine issues that apparently concern them, because, to paraphrase Gertrude Stein's description of Oakland, "there isn't any `they' there." Instead, such groups can be used to design product or service "lines" or to formulate arrays of options that seem complete-covering most of the considerations. Also, these groups can be used to define and map the factions implied by a number of related specific proposals for action. Panels constituted in this way are used all too frequently. As an alternative, consider making up panels that have two or three distinctive subgroups and manage feedback selectively.
Occurrences are what goes on with groups where:
All panels on occasion become forums to prescribe for the ills of society. These groups virtually insist on this mode. In most instances, this is counterproductive. This is because in most circumstances it is necessary to produce a fine-grained, narrowly bracketed reality to support specific actions. The reason for arranging this kind of a panel is to obtain a broad consensus, but not on everything. Since nothing is ever done in general, but only in a particular instance, introduction of situation-specific facts and considerations can help to focus the panel on agreeing about society's interest in the instance at hand, instead of on society's interest generally in things of this kind.
Another reason to arrange a broadly based general panel is to develop a contextual mapping that would describe the overlapping large-scale realities which underlie different parts of a society's response to any complex issue. Here it is useful to inquire about "for instances" that are likely to be seen by different elements of society as analogous to the situation being inquired into. "Well, it's just like thus-and-so... " and similar phrases are powerful connectives to ready-made meanings for a proposed idea or approach. These loose, even unrational, associations tend to bring along existing emotional loadings which panelists in this mode tend to deny. In a similar way, apparent metaphors (e.g., like David and Goliath) and apt personifications (e.g., rape of the ecology) should also be explored. This can be done by asking about which ones suggest themselves or provide several examples to be selected from, modified, or added to. Identifying analogies and metaphors is a more fruitful way to define subtle differences in global constructs than asking individuals to examine their core values and beliefs and describe them directly. While there are problems with interpretation of these complex symbols, the interpretations can be iterated for refinement, and also for validation. Since the "picture" is changing, and "observation" using a Delphi inquiry is not a snapshot, there is a gross limit on the fineness of the picture of a whole society that can be produced. But, even faint shadows on pale gray are helpful in many situations.
Do not do a Delphi with representatives or spokesmen for society when you want information for particular practitioners; use experts "about society" who are respected by the practitioners. In fact, you may want a split panel containing both experts and practitioners to create a joint reality shared by both that can be potentially extended to society.
What I Meant to Say
Most of the foregoing is tentative and much of it highly abstract. It is even embryonic. I share it for whatever use you can make of it and apologize for its density. There are obviously many hypotheses here for research on the process of Delphi interaction that go well beyond the issue of "accuracy" or "reliability." My only suggestion is that you try to supply or create examples from your own experience that seem to explicate these generalities. I attempt to summarize the main points below. In the next-and final-part of this article, I will offer a few pragmatic suggestions for the designer-monitor of Delphi inquiries. Obviously, not all of these strategies and techniques should be attempted simultaneously.
(1) Each Delphi interaction produces a shared reality which is initially formulated by the panelists from their expectations and the style of presentation used in the initial materials; this particular reality is elaborated and modified by the succeeding interactions.
(2) During the process of interaction the panelists' responses can be expected to deal with personal esteem, group self-concept, and relevant world-views as well as to convey substantive ideas, forecasts, and estimates.
(3) Patterns of styles of interaction can be fostered, retarded, even transformed in order to produce results that will have greater insight, more usefulness, higher impact.
(4) The size and shape of the reality within which things come to be viewed is more important than the specific substantive descriptions produced by the panelists.
(5) The believability and significance to the user of the results of a Delphi inquiry depend as much on the user's perception of the clarity, compellingness, and fit of the reality implied and possibly defined by the results as on the perceived quality of the information.
Some Design Considerations
I shall now briefly focus on some design considerations to help you develop a better Delphi and orchestrate the interaction. These suggestions do not constitute a checklist. I hope they will trigger some ideas and preparations that otherwise would not have occurred to you. Each could separately be the subject of a much longer discussion. Supplying these discussions is left to you, the reader, when you pursue particular applications.
Making a Delphi Context Specific
The conceptual productiveness of the interplay of intentions and circumstances is lost in many Delphi processes because of the lack of concreteness in the context. The mental life of most individuals involves attribution of an intimately known context, even when this context is the world of abstract ideas. However, the subjects of most inquiries, including Delphis, are stated in terms of general propositions; those things that apply only to an instance are defined as being of no significance for action (those things yet to be determined, decided, or done). So the difficulty lies in the extraction of generalizable propositions from particular instances.
I can offer two approaches; there are undoubtedly more. The first approach is to develop a concrete example with detailed features that are typical of a more general case. Another source of specific contexts is to create hypothetical constructs which make a distinct break with most present circumstances and are almost fanciful, e.g., on another planet. Because the hypothetical construct will differ in important respects from the panel's conventional circumstances, it can often provoke new thinking, retard discussions of routine experiences that are rooted in presently accepted and unquestioned conventions, and create an unclaimed and common experience base that can facilitate collaborative (as. competitive or compliant) interaction. Emergencies and other kinds of special circumstances can also be used as hypothetical constructs within which possibilities and insights can be developed for later transfer to the more general everyday circumstances.
Once a case is selected, proceed with the Delphi, focusing on the detailed case initially. Then shift the inquiry by asking the panel to create generalizations-possibly proposing some yourself and iterating the ones they proposed. You can next elaborate these general propositions by asking,"What other circumstances do you know or can you imagine might prevail, and within each would these propositions be valid?" The second approach is to develop several special cases for initial inquiry using subpanels. Later, the results of each subpanel can be shared with the whole. Then, ask for general propositions. What is forthcoming can then be refined and elaborated with examples.
The Domain of Time as a Context
Another shortcoming of "looking at things in general" is that time is assumed as an undifferentiated flow -- that one minute or hour is as good as any other. If asked, we all would acknowledge perceptual differences-time dragging or racing. But, unless our attention is specifically directed to it, we tend to overlook structural distinctions in time. The structural features that limit the allocation of time are of increasing importance as time becomes a scarce resource. When inquiring "What will people do?" "How can resources be used?" "Which group is being designed for?" the time domains become critical concepts. Time domains can be categorized or partitioned in many different ways, as can any continuum. Conventions in the dominant culture have created workdays (and, increasingly, work evenings and nights), regular times off, lunches, weekends, holiday weekends (recently added to by Congressional action), annual vacation times, "the holidays," special events, and so forth. There are also times in the so-called life cycle-youth, teenage, young married. Time domains can be treated as the subject of policy and design, for example, staggered work schedules and new work patterns (4 ten-hour days, 6 twelvehour days with a week off, etc.) You can also invent new "times" for things, like the enculturation of daily exercise periods. Figure 8, below, deals with several of these considerations and relates them to the selection of market opportunities.
Demand for recreational services is limited more by the time consumers have available for their pursuit than by the expenses involved. Time is experienced in periods that have each come to have their own meaning and expectations. To create new markets for recreational services, the meanings of particular periods or domains can be altered and new periods can be created and given meaning.
Round 1 : Suggest new time domainsyou think could be developed and indicate how existing domains could be revised or differentiated to create new markets for recreationnal services.
Round 2: Identify tune domains where additional involvements for working adults without children would be most welcome and suggest recreational pursuits.
Round 3: Select market opportunities you believe are the most attractive and suggest specific ypes of products and services and actions to create expectations and aggregate markets.
Getting a Product Out of Results
Every Delphi inquiry can be expected to produce results of some kind. Usually, if they are thought about at all, results are seen as things to be "captured" from their existence "out there somewhere" by the Delphi process. Instead, try to visualize, or think about, the results of a Delphi as being created by the interaction. just as the important properties of a building are not simply those of the sum of its "stones," the results of a Delphi are not just the individual items produced in the interaction but the reality comprised by the whole. I believe that a Delphi will be more productive if the monitor sees his role as producing results and not as "surveying" things that are already there.
Here are some more design tips:
Creating panels is usually the first task, unless they are a given-as with an organization that wants to study its own future in a participating manner. Even then you may want to augment the group. Three kinds of panelists are ingredients for creating a successful mix: stakeholders, those who are or will be directly affected; experts, those who have an applicable specialty or relevant experience; and facilitators, those who have skills in clarifying, organizing, synthesizing, stimulating... plus, when it seems appropriate, individuals who can supply alternative global views of the culture and society. The proportion of a panel from each category should be tailored for each application. Note: There are almost never enough women on panels.
There are no general rules of thumb for creating panels. For example, where options and interests are clear but acceptance of direction and action is fractionated, stakeholders might predominate. If it is clear who has to act, but not clear how, a heavy salting of experts may be best. Where issues, relationships, and values are unclear, a preponderance of facilitators may prove most useful. Pay attention to the minority as well as the majority views expressed by each type of panelist. Also, watch that panelists contribute as you expect--experts, in particular, drift into acting as stakeholders in aggressive moments and facilitators in passive ones.
After abstractly thinking about the panel, it comes down to names. Most of the time you will not know everyone you want. Sometimes you can start with a small group of potential panelists and begin by discussing possible names and searching as a group for interesting and appropriate candidates. This is a better strategy than searching lists of relatively unknown names by categories. Frequently, staff members of professional societies and other "people brokers" can be utilized to suggest panelists. Often the most fruitful part of a Delphi process is assembling a panel, not simply for the particular effort, but because it enhances your contacts. At least, try to develop two alternative tactics for identifying panelists and consider their trade-offs before embarking on the Delphi.
Stimulating response of any kind from quality panelists is not easy. Getting quality inputs is even harder. Few people like "questionnaires." And, from their point of view, engaging in abstract speculation with people you do not know or whom you want to impress about a subject that is not central to your interest (and where your performance is unrelated to your survival) is hardly compelling, Motivation requires more than a good cause, a pep talk, and thanks. If prospective panelists tend to be uninterested, find a "worthy" or prestigious sponsor, or make participation of significant publicity value. Then a token payment or "honorarium" will stimulate interested responses. Payments alone, even of fairly sizable sums, will not assure quality participation. It is difficult to pay panelists based on performance since it is difficult to agree on what is expected; panelists generally believe that "what I do is what's enough." Attractive and potentially stimulating peers are probably the most powerful incentive for participation. Consider gifts and "in kind" rewards for participants, because often a sponsor can provide these at modest cost-particularly travel. Still another way to organize participants is to employ a two-step approach: suggest that key participants use their staff, students, or others they can induce to participate to formulate the specific responses. The key participant then collects, reviews, and submits them for each round. Here publicity and recognition of importance by other individuals are powerful incentives.
Once initial participation is secured, the next important consideration is the quality of the materials, their tone, style, and presentation. Use lots of color. Give the materials style. What you send out reflects the significance of the inquiry and the value that is placed on it. Use emotive language and vernacular expressions to engage panelists and convey importance of results-not another abstract study. Detailed situational descriptions, mini-scenarios, bits of conversation-all help. Other media besides print can be used effectively to make response easier and more scintillating -- even speedier. These include tape cassettes and even local "interviewers" in cases where the Delphi sponsor or manager has a large potential manpower pool that could be employed at low cost. Don't forget to pay attention to creating an audience for the results and communicating this to panelists. Examples of good and bad responses are also useful, particularly for discouraging stereotypical remarks and obvious but unhelpful "insights."
Orchestrating interaction requires attention to details in the panelists' responses and a feedback of an overall movement, countervailing forces, or whatever macro-observations are appropriate to describe what seems to be going on between and with the individual items. Highlight divergence and consensus, even when both are true for different sets of responses to a quirk item. $e sure to cheer at least one item from each panelist-everyone needs to be appreciated. Introduce ambiguous factors if discourse focuses on individual items when you would rather explore relationships between items. You can also "stipulate" certain constraints when particular items have evoked massive, but trivial discussions of immaterial distinctions. Where appropriate, supply tentative theoretical constructs and cosmological frameworks. Indicate the way responses are being categorized for review and whether your interest centers on enriching the decision environment or identifying and refining options for consideration. Also, you can employ crass-impact analysis (see Chapter V), comparisons with analogous situations, and other techniques to provide a basis for evaluating responses in a different light. Again, at the risk of overemphasis in this essay, as interaction progresses, share with the panelists the reality construct(s) you believe they are referring to and shaping by their responses.
Interpretation and summation of responses is never complete, because the panelists do not send in their "heads" but only their responses. A complete record of the interactions is unwieldy and diffuse and difficult to use. It is important to begin "interpreting" responses during the interactions, even at the start. This makes interpretation subject to review by the panelists and can include their refinements, which I have found most insightful. Summation can be made from several points of view-using the one best for each major consideration in the inquiry. It is often useful to point out nil findings, omissions, and ignored items. Try to capture and describe the reality that was negotiated by the panelists and the monitor, because this provides a perspective for understanding and indicates the application foreseen for the results.
Communication of the results may not be felt by some to be a legitimate concern for researchers. Findings, it is supposed, have their own importance. However, the process of acquiring "understanding" carries with it an impetus for action that is not conveyed by presenting the conclusions alone. Most Delphi users, I believe, intend to influence the formulation of policy or the making of decisions. Doing this requires attention to the communication process. Well-organized, lucid, attractively presented reports help-but they are not enough. To augment the report you should attempt to: (1) create involvement--one way being to include the intended users or members of their staffs as panelists in the Delphi, or else to engage users in parallel deliberations that focus on the same issues, infusing from time to time interim considerations developed by the other Delphi panel; (2) generate interest-this usually means fanning respectable controversy and building a climate of expectation for, and awareness of, the Delphi inquiry among those who are important to the users, particularly colleagues and constituents; (3) present the results interactively-this can be done by organizing a simulation exercise where the panelists, the staff of the Delphi monitor, or some of both take roles that are significant in the user's operating environment and then enact a series of situations that can demonstrate which Delphi results are applicable, and when and how they might be used. It may be useful to define a communication process as part of the initial conceptualization of the Delphi design so that the implications for its communication are anticipated in the way the interaction of the panel is orchestrated.
I cannot imagine that all the recommendations in this Chapter will prove unassailable or always useful. I hope they will spark some interest and provoke a few insights. This will require a lot more of you than of me. I have benefited in putting this down, but you will have to work to make it apply to your situation in order to profit. To quote Henry James: "Many things have to be said obscurely before they can be said clearly."
1 Monitor is a term I use for the person or group conducting the Delphi inquiry, i.e., preparing the materials, interpreting the responses, integrating the insights, presenting the results, etc.
2 For a longer exposition of, and details about, this concept see S. Lyman and M. B. Scott, A Sociology of the Absurd, Appleton-Century-Crofts New York, 1970, and H. N. Lee, Percepts, Concepts, and Theoretical Knowledge, Memphis State Univcrsity press, Memphis 1973.
3 Edmund Husserl, Ideas: General Introduaton into phenomenology (traps. W. R. Boyce), Allen & Unwin, London, 1931.
4 Remy C. Kant, Merleau-Ponty and Phenomenology, in Phenomenology (Kockelmans, ed.), Doubleday, Garden City, N. Y., 1967.
5 Calvin O. Schrag, Phenomenology, Ontology and History in the Philosophy of Martin Heldegger, Revue Internationale de Philosophie, vol 2 (1858).
6 Jack Douglas, Understanding Everyday Life, Aldine, New York, 1970.
7 Hans Peter Drictzcl, Recent Sociology No. 2-Patterns of Communicative Behavior, Macmillan, Nev York, 1970.
8 Example: "Suppose you had invented tire better mousetrap and people had beaten a path toyour door-would they buy?"
9 Example: the famous rejoinder, "And how does that relate to the Jewish question?" or Stephen Potter's functional equivalent, "…but what of the South?"
10 Georg Lukacs, History and Class Conciousness (trans. Rodney Livingstone), M.I.T. Press Cambridge, Mass., 1971.
11 Erwin Panofsky, Meaning in the Visual Arts, Doubleday, Garden City, N.Y., 1955.
12 Presenting a less-than-complete concept takes guts. To each panelist there will be "obvious" omissions. You will be severely criticized. But, I believe a stance of calculated naivete produces the best results, if you live through it.