Abstract
This article questions why sociology still lacks elaborate methods of scientific forecasting. Following a critical discussion of two main obstacles for sociological futures research – complexity and interaction – the deductive scenario development approach is introduced. Since scenario planning stresses plausibility not probability, it enables sociologists to create logical, not likely pictures of alternative futures. As a way of producing scientific results concerning future developments as well as testing scientific models, deductive scenario planning is labelled as sociological quasi-labs.
Scenario building has been a neglected issue of professional sociology. In the light of the long tradition of epistemological dispute in the social sciences, prominently acuminated in the Thomas Theorem, the reluctance of professional sociologists to address methods of forecasting and futures research is not surprising. Thomas’s proposition highlights a cardinal problem of sociological research, caused by the essential ability of the very object of research to interact with (and by this being altered by) the scientific findings, and reminds us of the constant threat of producing self- fulfilling (Merton) or suicidal prophecies (Andreski). Sadly, the silence of sociology cannot be the answer. Due to the fact that there is a public demand for scientific predictions, particularly with regard to new and growing uncertainties, the refusal of professional sociology to cater to these demands gives room to a forecasting industry, filled with pseudo-scientific charlatans (e.g. John Naisbitt or the inescapable Faith Popcorn) who do not care much for methodological standards. In sum: sociological theory about future developments faces important obstacles and simultaneously competes with commercial market criers of megatrends.
Since the 1960s and 1970s scenario techniques have become increasingly popular within economic planning units and are beginning to influence (scientific) political consulting too (Schulz, 2009: 291). I think that it might pay off to explore the potential of these techniques as sociological test fields or quasi-laboratories. Especially the deductive scenario development process, which starts with the identification of the overall framework by defining the most likely driving forces, can function as a test range for sociological theory building and verification of existing theories as well. Institutional theories, questioning the conditions for institutional stability (path dependencies) and circumstances and modes of institutional change (punctuated equilibrium models, incremental change), can be used to develop plausible concepts of future developments. Since scenario building stresses plausibility not probability, it can serve as a touchstone for the consistency of sociological models and for their range and limits of explanation alike. Instead of getting lost in scientific prophecy or trivial trend spotting, sociological scenario building may help developing testable sets of ‘if…then’ statements, which may not predict the future, but rather point to possible future developments. 1
In this article, the question why there is still no complete consensus on standards and methods of scientific forecasting among professional sociologists is addressed. This question is neither trivial nor irrelevant. Its relevance stems from the public demand for scientific prognoses. It is not trivial since forecasting in the social sciences faces serious epistemological obstacles. In the following I discuss the potential of deductive scenario planning to overcome these difficulties – at least partially.
The state of sociological futures research
Despite being a topic of professional sociological research since the early 1960s, Future Studies have yet to develop into an accepted and well-defined field of sociological analysis (Bell, 1996: 39). Although this does not mean that there are no Future Studies around (particularly notable are the periodical ‘State of the Future’ Reports of the Millennium Project), most researchers associated with these questions seem to agree with Slaughter’s remark ‘[one] should have done more, gone further’ (2002: 229). The neglect of reflections on the future and accordingly a lack of effort on developing appropriate models and methods can at least partly be explained by the dominance of business planning units, profit-oriented contract trend spotters and ‘pop futurist clones’ (Slaughter, 2002: 230). Evaluating a range of influential publications of for-profit trend research, Slaughter (1993b) complains about a general lack of theory, tailored results (fitting customers’ and/or readers’ expectations) and normative bias of these forecasting efforts. The competition with the trend spotting industry is especially problematic for academic researchers because there is a debate whether sociological Future Studies should aim at possible futures (Anheier and Katz, 2009), or desirable futures (Bell, 1996; Masini, 2002). 2 The dispute is closely related to the question, if there is a future to discover, rather than a future to be shaped. Reminding us of the interactive nature of the future, Bell addresses the problem of self-altering prophecies, which contradict the classical positivist view of the science–object relation (Bell, 1996: 45). Taking up some post-positivists warnings of an ‘uncritical acceptance of the results of modern science’ (Bell, 1996: 44), Bell favours ‘critical realism’ demanding reasonable arguments for scientific assumptions rather than irreproachable proof (Bell, 1996). An opposite view is pronounced by Bueno de Mesquita, whose ‘expected utility theory’ aims at developing highly complex models for predicting policy choices (1988, 2011). Adopting the core principles of rational choice theory, Bueno de Mesquita develops models of iterative games, including changes of the players’ expectations by the game itself (2011). As convincing as the model is, as problematic are the assumptions derived from rational choice theory and the neglect of the possibility of reciprocal effects between the forecasting model and the strategic choices of actors involved (cf. Auinger, 1995). A more balanced framework was brought forward by Slaughter. His concept of a ‘knowledge base of future studies’ (1996) is a blueprint for combining theories, models and methods rather than outlining a ‘monolithic entity’ (Slaughter, 1996: 806). In the context of my article Slaughter’s pointing towards a main skid for futures research is especially important: the lack of a sound body of theory and methodology. The most ambitious attempt to remedy this deficit is the textbook on Futures Research Methodology by Glenn and Gordon (2009). The collection provides information on a variety of methods (including classical approaches like the Delphi method, scenario planning, mathematical methods like cross-impact analysis, but also a discussion of ‘soft’ approaches like ‘genius forecasting’ and environmental scanning). The compilation also provides some very useful thoughts for combining specific methods of futures research. Most important: the futures research methodology must be understood as a field report, rather than a project already completed (Gordon and Glenn, 2009: 12). This is where this article comes in. Focusing on one specific method, my aim is to carve out links between deductive scenario planning and the wider theoretical landscape of sociology. Of particular importance are the epistemological limits and opportunities of this approach. Sharing some basic assumptions of Bell’s critical realism, the ultimate goal of this article is to contribute to the ‘continuous process’ (Slaughter, 1996: 806) of strengthening the knowledge base of future studies.
Sociological futures research: Epistemological obstacles
While forecasting is prevalent in the business world, the question if and how predictions can be based on a resilient sociological methodology has always been a contested issue. Although a detailed presentation, critical discussion of the methodological conflict would go beyond the scope of this article, but two fundamental lines of conflict can be marked and emphasized.
First: the question whether the limitations of sociological forecasting derive from inefficient or underdeveloped theories and models about society and the process of social change. Here the problem of the complexity of social reality is addressed.
Second: the problem of interaction effects between prognoses and the people addressed prominently summed up in the concept of self-fulfilling prophecies.
The problem of complexity
… after all, physics, complex as it may be, is relatively simple as compared to a subject which includes physicists and physics, and everything else mankind has ever said and done on earth. (Beard, cited in Andreski, 1964: 38)
The first major concern for sociological futures research is the complexity of social reality, which includes natural and cultural phenomena. Sociological diagnoses let alone prognoses have to deal with highly interdependent processes, interaction between various and different actors and the influence of elaborate institutional frameworks. A practical way to address the problem of complexity is to limit the range and depth of scientific forecasting. While predictions tend to be less accurate with the covered span of time increasing, ‘general trends’ should (at least in principle) be better to foretell than individual events. Or, in the words of Young: One would not expect a social scientist … to be able to guess any more sagely than anyone else what De Gaulle or Ho Chi-Minh is going to do next month. Nor would he necessarily be anymore spry about particular events. His skill in dealing with facts relating to communities which are capable of being expressed by numbers is what provides him with his framework for prediction. (Young, 1968: 13)
The ability to predict future developments derives from the ability to deal with countable quantities and mathematical provable connections. Young’s 40-year-old view represents a still timely position. The better the (mathematical) skills of (macro) sociology to pinpoint general linkages of countable quantities, the better the ability to predict future developments (cf. Bueno de Mesquita, 1984, 2011). The problem of complexity seems to be reducible to the question of how to improve the instruments of statistical analysis. Although an advancement of statistical techniques would always be very valuable, it cannot solve the problem of complexity, for two reasons. First and foremost there is no logical necessity for every relevant aspect of social reality to be measurable (in a mathematical sense). Furthermore, even where mathematical connections can be verified, the risk of producing statistical coincidences remains. Auinger, for example, reminds us that mathematical laws in the social sciences often tend to confuse the possibility of a connection with necessity (Auinger, 1995: 179). 3
The problem of interaction
A second and even more substantial obstacle to sociological futures research stems from the fact that social beliefs can and do shape social realities. Since such beliefs (that means the knowledge which is being used to make sense of social circumstances) always need to be legitimated, sociological findings can influence the very way how people make sense of the social reality they live in. People are not only parts of overall developments but can take part in shaping future developments (Bell, 1996: 45). Predictions therefore can encourage and even persuade (maybe because their actions seem to be in line with some general trends which are scientifically ‘proven’) or dishearten people (because they are not). Sociological description may end up in (unintended) prescription: ‘Human beings not only react to being observed, but are also influenced by what is said about what will happen to them, or about what they themselves will do’ (Andreski, 1964: 38).
This thought was prominently formulated in the Thomas Theorem: ‘If men define situations as real, they are real in their consequences.’ The problem of interaction was also the topic of Robert Merton’s famous work about the self-fulfilling prophecy (Merton, 1957). This concept refers to the instance that scientific statements can alter the related reality in a way which subsequently leads to their correctness. Merton constantly reminds us of the special relation between the researcher and his or her object of research in the field of social sciences. Otherwise such statements can make sure that they will be wrong in either case, by influencing people to avert predicted developments (Andreski, 1972). In both cases, the fundamental ability of the object of research to interact with the results of research (in fact with the researcher herself/himself) is an integral part of the sociological research conditions. In contrast to the natural sciences, the sociological object of research ‘cares’ for the result of professional studies, especially when likely developments, actions and trends are concerned. Since sociological diagnoses and prognoses can affect (or even alter) the definition of a situation and therefore affect (or even alter) the consequences, the improvement of analytical methods alone will not solve the problem of scientific forecasting.
No laboratory conditions in the social sciences
A physicist or a chemist, when he finds an explanation unconvincing, can usually check it by repeating the relevant experiment. (Andreski, 1964: 40)
A textbook definition of ‘lab conditions’ would stress the controlled conditions which allow the verification or falsification of research hypotheses. ‘Controlled conditions’ refers to the instance that the research process is repeatable and the results reproducible (at least in principle). In the social sciences, repeatability is limited to specific statistical methods as long as the data basis remains the same. The process of data collection is normally not repeatable – be it a problem of feasibility (in the case of large data banks) or a problem of quality (in the case of repeated questioning, which can induce ‘learning effects’ of the respondents) – which limits the control of the researcher. Even more important: while in the natural sciences laboratory experiments aim at uncovering stable interactions (laws), sociological research always depends on the adequate interpretation of a given situation (Weber, 1978: 11). Discussing the potential use of a (hypothetical) system of exact social laws, Weber concludes that such a scheme could merely be a useful tool, a first step of sociological explanation: The determination of those (hypothetical) ‘laws’ and ‘factors’ would in any case only be the first of the many operations which would lead us to the desired type of knowledge. The analysis of the historically given individual configuration of those ‘factors’ and their significant concrete interaction, conditioned by their historical context and especially the rendering intelligible of the basis and type of this significance would be the next task to be achieved. (Weber, 1949: 75–76)
Weber’s definition of laws in the social sciences equals a rejection of positivists’ (who search for universal laws in the social sciences) and subjectivists’ (who aim at the immediate experience of individual situations) positions alike. He distinguishes between sciences which aim at explaining the concrete reality and sciences which try to discover universal laws (cf. Hekman, 1983: 155ff.). Among others, sociology belongs to the first category, which means that the cultural significance of a given research question is emphasized (Bell, 1996: 44). Scenario planning as a sociological approach therefore must not be directed towards discovering universal laws or trends.
I want to argue that deductive scenario development may offer an alternative to the problem of interaction and comply with the Weberian view of social sciences. Because scenario planning aims at plausibility, not probability (and is therefore compliant with Bell’s [1996] critical realism approach), this method of theory-based story telling may lead to credible narratives of future events without prescribing the things to come or formulating misleading systems of apparently universal laws. Simultaneously, deductive scenarios can provide frameworks within which data can be conceptualized and hypotheses can be tested. It will be argued, therefore, that deductive scenario development can function as sociological quasi-labs.
Towards sociological quasi-labs: What deductive scenario planning is, and what it is not
‘Scenario building is a narrative forecast that describes potential courses of events and actions’ (Anheier and Katz, 2009: 244). The development of future scenarios is one – very specific – way of scientific forecasting. Broadly speaking, scientific forecasting deals with the problem of how to derive lessons for the future from scientific findings. The touchstone for the quality of a scenario is, as mentioned before, plausibility not probability. By this, the methodological approach guarantees an awareness of the danger to prescribe the future by describing the content of a prognosis.
Box thinking!
Scenario planning facilitates a structured approach. The deductive scenario development process starts with the identification of the overall framework by defining the most likely driving forces. Those drivers include general economic, cultural, social or technological developments. The identification of key drivers draws heavily on scientific diagnoses and theoretical models of social change. The dominant drivers mark out a continuum of high and low influence which can be illustrated as a simple coordinate system which frames two more extreme (Scenario II and III) and two rather ambivalent (Scenario I and IV) development paths (see Figure 1).
Drivers and scenarios.
An obvious reason to criticize this approach is the rigidity and arbitrariness of the selected drivers. In fact, due to the major problem that the overall framework for deductive scenario development relies on the assessment of likely and plausible drivers, it means that any scenario is the product of a limited and preliminary state of knowledge. (Slaughter, 1996: 810).
To increase the plausibility of the scenario setting, the method makes use of theoretical concepts which help to canalize change and by this limit the scope of uncertainty. Simply put, (lasting) social change translates into new institutional arrangements. It is therefore plausible to rely on existing theories of institutional change to identify key drivers, uncertainties and opportunities for change. The single most important concept of institutional influence of social change is the thesis of path dependency. In combination with theories of social change a framework can be developed which helps to derive the relevant questions for the causation, direction and modes of social change.
Theory-based story telling: Implications of institutional theory for scenario planning
Thinking about future developments implies sound knowledge about the past and a precise analysis of the present situation. To link knowledge of the past and insights into current developments for scenario planning, sociological theories of social stability and social change must be taken into account. I want to argue here, that institutional theories provide a particular suitable framework for scenario planning.
Path dependencies as guidelines for scenario planning
A common assumption of institutional theory is the thesis of path dependency. Social change is in this sense channelled by existing institutional arrangements (‘history matters’). Although fundamental change is not ruled out completely, path dependency stresses mechanisms which tend to stabilize institutional development paths (cf. Beyer, 2006). Scenario planning can use concepts of path dependency for modelling plausible developments on the one hand. On the other hand, the scenario technique can be used to detect possible points of path departure or even path ends (under which circumstances is a path-dependent progress no longer plausible?).
Modes of institutional change: Crossroads of scenario planning
Institutional theory, especially the new institutionalism in sociology, tends to emphasize institutional continuity. Institutional change is often the ‘Achilles heel of new institutionalism’ (Harty, 2005: 52). It is therefore important to consider two fundamental modes of institutional change. They indicate sudden or gradual departures of established institutional pathways and so mark potential crossroads in the scenario planning process.
Incremental change: Primacy of institutions
The first ideal type of institutional change is incrementalism. Change is modelled as sequences of small logical steps (Dunphy and Stace, 1988: 318). It is described as long-term adjustments according to (existing) institutional settings. In the scenario planning process, the model of incremental change can be used to consequently think ahead on the basis of present institutional settings.
Punctuated equilibrium: Nothing stays the same
‘Punctuated equilibrium theory seeks to explain a simple observation: political processes are generally characterized by stability and incrementalism, but occasionally they produce large-scale departures from the past’ (True et al., 1999: 115), or ‘game-changing shocks’ (Bueno de Mesquita, 2010). For the scenario planning process is it important to identify the conditions for such critical junctures to take place. Nevertheless, a major problem remains. Since institutions seem to explain everything (path) until they explain nothing any more (that is a break within the ‘historical logic’ takes place; Blass, 2003: 1050), it must be clarified in the scenario development process which new arrangements will (plausibly) form the a new equilibrium.
Pitfalls: ATAMO and unique occasions
There are many obstacles and dangers for the scenario planning process. Although every development process faces individual troubles, two major pitfalls for the scenario builder must be mentioned.
ATAMO (and that a miracle occurs)
ATAMO is an acronym for explanations which apparently illustrate the course of a given scenario but are not justified themselves. A prominent example would be the change of consciousness: change happens just because people realize that they have to act, behave and think in new ways. The question remains why people should attain enlightenment! The suggestion that mysterious changes of consciousness or values drive social change conceals the complex interactions between actors and institutional settings. Change is then a consequence of a miracle occurring.
Unique occasions
Unique occasions often function as explanations for sudden and/or dramatic changes. According to the sociological model of the punctuated equilibrium such occasions represent moments when the traditional order is suspended and a window of opportunity opens up. Like ATAMO, unique occasions are able to explain lasting change in a scenario setting. Instead of leaning change on the back of rather vague changes in the consciousness of people, unique occasions seem to be promising because they clearly name the event which changes everything. Moreover, as history tells us, there are turning points and breaks! Relying on a unique occasion alone to explain the progress of a given scenario includes the danger of overemphasizing the influence of punctuated equilibriums. The problem with a model which describes change as a reaction to unique occasions in the first place is of course the question whether or not change can happen without such an occasion. Scenarios are more vulnerable the more they (solely) depend on unique occasions to explain specific developments.
Quasi-lab conditions
The deductive scenario development process outlined above complies with the conditions which qualifies it to function as a sociological ‘quasi-lab’. Scenario planning can be labelled as a test field for sociological modelling, since it confronts theoretical concepts with hypothetical developments. Constantly judged against the touchstone of plausibility, these concepts can be pushed to their limits. Since social change is necessarily a complex phenomenon, in which strategies, counter-strategies, opportunities and sometimes simply chance interplay, the aim of such scenario labs cannot be the generalization of crucial law-like mechanisms. Complex developments must be unravelled in rather simple sets of ‘if…then’ questions. The main goal of scenario planning is the identification of the conditions and circumstances under which predicted developments are no longer plausible.
To describe scenarios as sociological quasi-labs includes an aspect of accentuation. On the one hand, the similarities to laboratories can be emphasized: scenarios function as quasi-labs because the coherence of sociological explanatory models can be tested with selected framework conditions kept constant. This can help to advance old or develop new models of explanation: ‘The purpose of science, either natural or social, is explanation not prediction. But prediction is one of the best means of testing explanation’ (Young, 1968: 20).
There are fundamental differences between quasi-labs and real laboratory conditions on the other hand: following (for example) the common objection that precise modelling of future developments is impossible for social scientists, mainly because of fundamental uncertainties regarding the timing of future events, scenario planning is limited to describing the inner logic of event chains. Again: scenario planning aims at plausibility, not probability. Scenarios can be contaminated by wrong assumptions or blind spots which may affect the results. A central question which tells if a given scenario planning process qualifies for lab conditions is, as noted above, the repeatability of the process. A touchstone for the plausibility of a scenario is the question whether or not the scenario would (all other things being equal) be constructed the same way again.
Conclusion and perspectives
Deductive scenario planning provides a framework for the integration of existing sociological theories and approaches to develop plausible narratives of future developments. It therefore has the potential to bridge the gap between the public demand for scientific prognoses and the scientific standards which are often violated by representatives of the fashionable trend spotting industry (Slaughter, 2002). Simultaneously, scenarios can function as test fields for existing sociological models. Complex sets of ‘if…then’ statements can disclose limits and implications of theoretical considerations. Circumstances where plausible narratives are no longer possible by relying on existing models alone highlight new research questions, or the necessity for additional theory building.
The systematic use of scenario planning in the social sciences has yet to be fully implemented. Especially the interplay of existing sociological methodology and scenario planning has to be examined more thoroughly. As was argued here it may pay off to think of scenarios as quasi-labs. As a bidirectional method, scenarios may enable scientists to make plausible statements about future developments without making prophecies. The method is thus compliant with Slaughter’s call for ‘a strategic view, to explore options and alternatives, to anticipate eventualities and prepare for contingencies’ (1993a: 291) as a baseline for future studies. At the same time the scenario development process may help to get a better understanding of the advantages, limits and gaps of existing sociological models for social change. Scenario planning is therefore a promising approach in regard to producing scientific results as well as testing scientific models.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.