Abstract
Generalization in research takes the form of statements where knowledge is claimed to be valid for objects beyond those actually studied. Can action research knowledge, with its emphasis on cooperation between research and those concerned, apply this notion? Several action researchers seem to answer this question in the negative in the sense that they introduce specific measures to support the broad application of knowledge emanating from action research, such as locating projects within networks, working with leaders who can carry the knowledge to further users, organizing the project as large scale events, and more. The purpose of this article is to discuss some of these measures with a view to bringing them together in a broader perspective on how to make action research knowledge reach out in society, with participative constructivism as the core concept.
How to make knowledge general
In the introduction to “Handbook of Action Research,” Wicks, Reason, and Bradbury (2008) see action research as inquiry grounded in the ecology of life, characterized by cooperation between research and those concerned, to promote emancipation and a better life. In spite of claims to qualities of this kind action research is, in most countries, a marginal activity and even where action research enjoys a reasonable popularity, it plays no major role in research policy or national innovation strategies. In a discussion of this paradox, Greenwood (2002) argues that action research is actually fraught with “unmet promises and unfulfilled challenges.” Action research needs to seek a re-establishment of the high profile innovative cases of the era of Kurt Lewin and his disciples, to be able to provide the world with new examples and new inspirations. Against this, it can be argued that single cases, however splendid they may be, have rarely diffused to other actors, nor seldom given rise to broad change. In fact, if they had done so, the world would already, after more than half a century of action research, have been shaped according to the ideals of action research. But if outstanding cases cannot be expected, more or less by automatism, to create broad change, what is then to be done?
This challenge has sometimes appeared as an element in specific projects, in the form of efforts to go beyond the single case to achieve a broader impact as a part of the project itself. The probably most common approach is to rely on the power of “the first project” to immediately convince new practical users. Making it possible to understand the project through its practical manifestations, action research has the advantage of being able to talk directly to new practitioners. Through attracting a growing number of users, the project can generate a multiplication of projects until “critical mass,” and, with this, self-sustaining change are achieved (Emery & Thorsrud, 1976). The potential of this approach is strengthened if the project site is part of a broader network that can help transmit impulses to other sites and it will be a point in itself to find action sites that are embedded in networks, or where surrounding networks can be created as part of the project (Chisholm, 1998). It is also possible to strengthen the diffusion potential of “the first project” through designing it in such a way that the conditions for broader use are built into the project design from the start (Engelstad & Ødegaard, 1979). A further possibility is to concentrate on leaders as project partners and rely on them to transmit the message to all members of the organizations they lead (Torbert, 2001). The leaders can be of many different kinds, ranging from group supervisors to world leaders. There is also the possibility of organizing action research projects as large scale events, to include as many people as possible from the start (Martin, 2001). There may be other possibilities. The point that there are differences between generating cases on the one hand and creating broad change on the other is, on a more general level, emphasized in the distinction between second and third person inquiry (Torbert, 1998). While exemplary cases can generally be created in face-to-face encounters, to reach out in society in general, there is a need for processes that can involve actors who have no direct relationship to each other.
From discussing how action research can be shaped in such a way that it is able to reach out broadly in society, one may move further to include discussions under such headings as networks, clusters, (learning) regions, innovation systems, social movements, and more. Discussions under these headings can include issues and concerns of relevance to action research without, however, having the challenges of action research as a major focal point.
Using experience from action research-based efforts at diffusing the idea of autonomy in work in the Scandinavian countries, in particular Norway, the purpose of this article is to discuss how knowledge generated in action research projects can be used to promote broad change. For about half a century—from the middle 1900s to the early 2000s—action research participated in a number of efforts to promote autonomy in work. Today, the Scandinavian countries—here taken to imply Denmark, Finland, Norway, and Sweden—show the highest scores on autonomy in the contemporary European working conditions surveys (Lorenz & Lundvall, 2011). It is often argued that the influence exerted by action research is owing to these countries being small, relatively free of conflict, and showing modest differences between people; characteristics that are presumed to make it easy for action research to make itself felt in society. This is, of course, possible, but it is also possible that the influence is owing to the way in which action research has developed and performed its role. In fact, looking at the history of the Scandinavian countries, it will be seen that when entering the process of industrialization, there was no lack of hierarchies. Nor was there a lack of conflict: in the period between the world wars, Sweden as well as Norway was, at times, on the top in the European statistics on days lost in labor conflicts (Gustavsen & Hunnius, 1981). The size may be modest compared to, say, Germany or the USA in general, but corresponds fairly closely to the average size of the 200 or so nation states that constitute the world community, as well as to many of the regions, states, and similar that often appear as natural areas for work reform even in the countries with large populations. It is, consequently, possible to argue that when the Scandinavian countries today appear as peaceful, co-operative, and even innovative, it is because of conscious acts performed by human actors and that action research can be found among these actors.
How to make knowledge reach many people overlaps with the issue of generalization of research knowledge. In descriptive-analytic research, generalizations beyond the phenomena actually studied take the form of claims expressed in texts. To what extent the textually expressed claims actually become known among those to which they pertain is of little concern; of even less concern is the extent to which they actually use the knowledge. Given such definitions as mentioned initially, this is hardly satisfactory for action research. It would be a paradox to assume that knowledge about, say, how to perform inquiry in cooperation between research and the people concerned can be diffused to new actors without cooperation. For action research, generalization becomes identical to the extent to which the knowledge is brought to influence human practices.
The point of departure
The idea of underpinning work reform with research-driven field experiments first came to Scandinavia through French who, together with Scandinavian colleagues, in the 1950s conducted an experiment with participation in an industrial plant (French, Israel, & Ås, 1960
Of the measures to achieve scope listed above, practically all were brought to bear on the diffusion of the experiences from the first experiments: employers and unions, both, launched information campaigns to strengthen the platform for practical pull. Both parties had, at the time, major training centers and much of the training was adapted to incorporate the experiences from the experiments. The leadership element was strongly present on many levels: the labor market parties centrally had leadership functions in relation to their members; on both sides, the top leaders in office at the time became involved; in the companies’ housing the experimental sites management as well as the union leadership was involved. Elements of large scale design could be found, in particular, as expressed through large conferences organized by the labor market parties separately and jointly. Action research was engaged in many of these activities, as advisors, speakers, and organizers.
In spite of all the efforts at diffusion, there emerged no wave of further projects that could be seen as multiplication of the first ones. Instead, from around 1970, there was a successive decline in the interest of unions and managers in performing projects along the same lines as the first experiments; a tendency that was discussed under the heading of “diffusion problems” and made subject to various interpretations (Bolweg, 1976; Gustavsen & Hunnius, 1981; Herbst, 1974). What continued, however, was a broad range of discourses about issues such as division of work, democracy, participation, labor-management cooperation, and more. Out of these discourses, there emerged new initiatives. Common to them is that the notion of autonomy in work was sustained but that major changes started to emerge concerning how to make the idea come real.
From expert driven to user driven change
When the experimental approaches withered away, they were replaced by other types of initiatives. A major step within the Norwegian context was a job design seminar organized by the labor market parties, but run by action research (Engelstad & Ødegaard, 1979). Originally designed to rationalize the diffusion of impulses from the experimental projects to new workplaces through inviting a number of organizations to participate in the same event, the seminar actually came to change the whole agenda for diffusion. When a number of organizations met in a shared context, they changed the program: the participants wanted to start the process with presenting themselves and their challenges to the other participants. As the next step, they preferred to explore the experiences with change and development that were made by the other participants. Only when these sources were explored did an interest in relating to external examples emerge. The core issue no longer appeared as the diffusion of experiences from a set of exemplary cases, but as the generation of interest in certain forms of organizational change. If the interest could be created, and the organizations pulled into a context where autonomy in work was the main topic, the working out of practical solutions did not present insurmountable challenges. In fact, some of the participants in the job design seminar developed, mostly on their own, projects that were as substantial as the first experiments. Several of the main automobile dealers—representatives of Volkswagen-Audi, Mercedes-Benz and Volvo—developed, for instance, forms of work organization in their bus and truck service workshops that implied organizing the mechanics in autonomous groups, each group with its own set of tools within their own group domain, and with its own set of customers. Dispatchers and foremen were mainly done away with. All experienced a significant decrease in errors and customer complaints. Since these projects relied more heavily than the first experiments on the ability of the users themselves to perform the job redesign tasks, the main action research contributions shifted toward the organization of the projects and the administration of the various discourse arenas that had to be created.
While the original experiments involved four worksites, and the various efforts to continue with the same project strategy added perhaps another 5–10, the job design seminar reached about 40 (Engelstad & Ødegaard, 1979). Some other projects emerged as well; altogether, however, the figures were modest in this period.
From design to conversation: Agreements on workplace development
When the labor market parties in Norway decided to support the first experiments, they set down a joint committee. In the early 1970s, this was turned into a permanent cooperation council. Given the problems associated with generating reforms on the basis of exemplary cases, the parties started, within the context of the regular renegotiations of their agreements, a discussion about a possible renewal of this apparatus. In 1982 a new agreement on workplace development appeared, introducing new perspectives and replacing the cooperation council with a new organization.
A main characteristic of this agreement is that it did not express views on criteria for “good organization.” Focus was, instead, on the processes needed to create this kind of phenomenon. The agreement, consequently, focused on how to make the local parties involve themselves in conversations on work organization. The main measure to be applied in this context was labor-management dialogue conferences. Action research was a partner in the discussions that led up to the agreement. The main role of research was to synthesize, and make explicit, such experiences as those from the job design seminar, to suggest possible measures to be applied in the implementation of the agreement, and to participate in conferences and other events organized under the agreement to ensure a process of continuous feedback to the labor market parties centrally, as well as a continuous improvement of the measures.
No dialogue is possible unless the participants accept the autonomy of the other participants from the start. Autonomy was, in this way, made into the crucial characteristic of the process of change from its starting point and not as something to be achieved after a more or less lengthy process of job redesign. The conference was constructed in terms of a set of criteria for participation and a set of criteria for the conduct of the encounter. It is dealt with in several publications (Gustavsen, 1992, 2001) and no further presentation will be done here. Basing the design of organizations on processes where autonomy is a chief characteristic from the beginning, autonomy is expected to characterize the outcome of the design process as well. No participant is expected to promote forms of work organization where the autonomy granted in the conversation is traded away in its outcome.
That the agreement struck a reasonable balance between outside initiatives and internal forces in the membership organizations could be seen from the fact that about 300 user-initiated conferences were organized during the 1980s. When all measures are considered, the agreement had, by 1990, reached about 500 companies (Gustavsen, 1993).
When the agreement was to be renegotiated in 1990, some observations came to play a role (Gustavsen, 1993): first, that relatively few of the agreement users had proceeded beyond organizing a conference. Of cases with more deep going change, around 30–40 could be identified. Second, there was little contact between organizations. In spite of the positive experiences with collaboration gained in the job design seminar, most enterprises tried to move on their own. Third, although research was involved, there was no organized framework for this involvement. The recognition of these points led to several steps throughout the 1990s. Major among these was the initiation of a new program called “Enterprise Development 2000” (ED 2000) that came to constitute the new setting for action research.
A distributive program
The main purpose of the program was to strengthen the context of workplace development, not only because previous experience had indicated that organizations could pull each other in processes of change, but also because of the recognition that the knowledge on which they relied was generally of a hybrid nature (Latour, 1987): it did not consist of a linear set of arguments and steps but rather of a mix of elements from different sources to be put together by the actors in each specific context. The points and arguments could come from different sources, and the sources need not to be outstanding examples. The job design seminar had demonstrated that a group of organizations could find ways in which to approach autonomy in work largely through drawing upon their own experiences without any of these experiences initially being of a “spearhead” nature.
When the program was to be concretized, a distributive pattern was chosen. Rather than base the program with one or two particularly qualified research groups, it was to be distributed on a number, spread all over the country. The idea was to start from many different local points, in the hope of making each point grow. It was thought that by emanating from local contexts, such forces as familiarity, regional perspectives, and similar could be used to attract participation. Seven units of researchers and companies were created in this way.
Within each unit, it was a preference for having several organizations participate from the beginning, not because the number of organizations was decisive for the mass of the project, but because it would make the participating organizations accustomed to working with other organizations from the start, as they had done in the job design seminar. If a local combination of organizations was successful, it was expected to be able to pull more organizations into the network.
As procedures were concerned, all units had to apply those that were agreed on as a point of departure for the program. This came, in particular, to pertain to democratic dialogue and dialogue conference, since these became the prime measures of the program. Otherwise, each research group was free to develop its own framework. When Marrewijk, Veenswijk, and Clegg (2010) criticize the notion of democratic dialogue for being an insufficient framework for organization development, they are right, but they have also misunderstood. Democratic dialogue is an institutionally anchored set of minimum critical conditions for cooperation, not a “full package for change.” Nor is it subject to change and revision purely on research grounds. Developed together with the labor market parties, the notion can be restructured only in cooperation with the parties.
The actual projects varied broadly, not least in terms of the ability to grow from small nodes to larger networks. In some of the cases, the formation of broad networks, or links to already emerging networks, occurred, in other cases the number of organizations in each unit remained limited. The more specific goals and patterns of the development processes likewise varied. In, for instance, the Nord-Vestforum network (Hanssen-Bauer, 2001), the main purpose was to extend a management training program into a program that could include broad participation from all employees in the membership organizations. In the Raufoss industrial cluster, 30 companies, that had formerly been one, set out to maintain the advantages of scale through working together, but in such a way that each could develop its own distinctive competence (Johnstad, 2007). Variation characterized, furthermore, the methods, or procedures, applied by research in the various local projects, as well as their theoretical anchoring. The anchoring ranged from strongly structuralist theory, to linguistic perspectives such as those argued by Pålshaugen (1998) according to which the only path to the understanding of organizations is through entering ongoing conversations with their members, at the same time as change is identical to various forms of interventions into these conversations. This program demonstrated clearly that there was no single best way of supporting organization development. Although the main emphasis was on organizing dialogues, there were also users who preferred expert advice, or even “lecturing” (a broad presentation of the program and its results can be found in Gustavsen, Finne, and Oscarsson (2001)).
On the basis of an evaluation (Bakke, 2001) the number of organizations showing significant internal changes in labor-management cooperation could be estimated to be around 40. Even though the total number was modest, an important aspect was that it had been growing throughout the program. Instead of the rise and fall curve that characterized the experimental projects of the 1960s and 1970s, the program strategy demonstrated a potential for continuous growth. A further point was that even if the number of organizations that had undergone major change was limited, there was, by the end of the program, a substantial number of organizations linked to the core ones through network formations (Claussen, 2001).
Revisions around 2000
At the turn of the Millenium, the agreement on development was renegotiated, so was the ED 2000 program. The first perspective to be pursued in this revision was to strengthen the development toward networks and other configurations including more than one organization. The second was to promote the emergence of a superstructure that could provide links between different networks, help initiate new networks, and generally fill functions in the terrain between and above the networks. At this time, the national government instituted partnerships to promote growth and innovation on a regional level. It was decided to link the ED 2000 program—now renamed into Value Creation 2010 (VC 2010) —to this development. This was made possible since the labor market parties were represented in all the 19 regional partnerships found in Norway. While the program continued to focus on autonomy and other quality of working life aspects, the notion of innovation as a goal was strengthened.
At this time, the agreement was gaining ground, in terms of number of users as well as in terms of impact in the user enterprises. By 2010 the estimate, as made by the employer-union board responsible for the implementation of all cooperation agreements, was that about 2000 companies had used the agreement and that about 500 had substantial results to show for the effort. Lacking the resources needed for thorough investigations of each case, the figures need to be taken as approximations and indicators rather than as exact measurements. This notwithstanding, there is little doubt that the agreement on development helped transform the movement toward autonomy in work from the level of dozens to the level of hundreds.
The institutional context
When the labor market parties in Norway decided to participate in the Industrial Democracy program, it was the beginning of a process that has been going on ever since, where the labor market parties have developed and restructured their cooperation and sent signals to working life concerning what the good work should look like, or concerning how it should be achieved. In the 1970s, this process came to include the state. The context was a new health and safety reform. Such reforms occurred in practically all industrialized countries during the 1970s and largely for the same reasons (see for instance Ashford, 1976; Bagnara, Misiti, & Wintersberger, 1978). While the traditional labor protection legislation had to some extent succeeded in combating accidents and dangerous substances in the workplaces, there emerged major shortcomings also within these areas. Equally important was, however, the recognition that there were major areas that were scarcely covered at all, such as ergonomics, the impacts of dysfunctional forms of work organization on issues ranging from stress to alienation, and interaction between different factors where each could be harmless but where the danger could be found in the combinations. If a reform is to be based on a strict principle of legalism—something can be regulated only when it can be fully specified—there was in fact no way in which the new themes could be included. In Scandinavia in general, and Norway in particular, there emerged, however, another way. If a health and safety reform could be based on labor-management cooperation, it would be possible to include all sorts of topics, since the criteria would be decided locally. To choose such an alternative became possible since there was an established tradition of cooperation that had manifested itself in local projects. They included only a minority of workplaces, but an amount sufficient to base the reform on co-operative achievements and expect others to be able to follow suit (Gustavsen, 1980; cfr. the notion of evolutionary, or reflexive, law as argued by Teubner, 1989). In this way, a health and safety reform came to function as a mechanism for the diffusion of autonomy in work. Action research performed three functions within the context of this reform: first, through having a member in the work group that developed the proposal, action research formulated a section in the act where some of the main characteristics of autonomous work were expressed (section 4 in the present Norwegian Work Environment Act). Second, within the work group and later, action research made substantial contributions to the procedures for implementation of the act, where a main point is to achieve broad mobilization and labor-management cooperation. Third, action research developed a new generation of field projects designed to illustrate how to implement the reform (Gustavsen & Hunnius, 1981).
The various steps taken by the labor market parties, in combination with such public efforts as the Work Environment Act, constitute an institutional setting for working life. In his overview of the socio-technical school in organization development, Trist (1981) remarks that this implied a legal sanctioning of the criteria for the good work emanating from humanist psychology (“psychological job requirements”). In one sense this is correct, in another it implies a misunderstanding. When criteria such as autonomy and learning are given institutional expressions, the reason is that “psychology” is in itself not strong enough as a platform for the actual promotion of autonomy. In particular, the unions have argued that if autonomy in work is a good thing, there should be few arguments against making this point in laws and agreements.
The impact of this reform on the development in work organization is difficult to assess. Various investigations performed during the first years after the law went into force indicated that it was mainly the larger industrial companies that had implemented the law in a way that had actual workplace impact, while other enterprises as well as the public sector were lagging behind (Gustavsen, 1983). There was a very high degree of overlap between those who used the agreement on development and those who implemented the autonomy-promoting parts of the health and safety reform. Over the years, the scope of the impact has grown, encompassing areas outside the joint areas of the main labor market parties, such as trade and service and the public sector. The growth has been sufficient to help locate Norway among the leading societies in Europe in the implementation of autonomy in work. This does, however, not mean that the law is fully implemented in all workplaces.
The common denominator: Participative constructivism
The main point in the somewhat lengthy and complex story recounted above is that the main contributions of action research refer to contexts as much as to cases. This emerges in particular from the shifts that occurred during the period. Among these, there are two that stand out particularly strongly. First, the one that became apparent around 1980 with the agreement on development. In this agreement, a major focus was placed on the social mechanisms that generate patterns of organization. Obviously, this was not a new recognition but rather a strengthening of a perspective that had been present all the way. It is not possible to perform any kind of action research project without conversations with those concerned: what happened around 1980 was that the main focus shifted from the topics of the conversations to the conversations as such. This shift can be likened to “the linguistic turn” that came to characterize much social research at the time, and there were obvious influences. As the work research discussed here is concerned, it is, however, important to emphasize that the shift was brought about by practical events and concerns rather than by a pure “paradigm shift.” With this, the core focus came to be directed at the conversations in which action research became involved: what were the characteristics of these conversations, what options did they give for action research, and how could they be made subject to influence?
Departing from the significance of conversations, the second shift that appears as major when looking at the whole period, but which is actually also a gradual one, is the growing emphasis on the relationships that surround the individual field site. The perspective was present in the Industrial Democracy Program of the 1960s in the sense that the status of the experimental sites in Norwegian working life in general was an important consideration in their selection. What has happened later can be seen as a process of differentiation: instead of thinking in terms of “one” surrounding network, today’s processes rely on distinctions between many different forms, such as networks, clusters, innovation systems, industrial districts, regions, interest organizations, social movements, society, and more. There is a broad literature on all of them and it would break the framework of one article to go into details. In brief, the notions of network and cluster generally refer to cooperation between organizations on a scale that presupposes direct contacts of some size and duration; innovation system often has a broader interpretation to cover larger slices of organizations that may benefit from the same overall framework without strong direct contacts between all (i.e. Silicon valley); the various notions of regions and industrial districts generally refer to mechanisms of integration within geographical areas of some size (well-known examples in the literature are the North Italian regions of Emilia-Romagna and Tuscany) and concepts such as the labor market parties, social movement, and society refer to processes where many actors or organizations move in the same direction but without direct relationships between the participants. Most enterprises will, at each and every time, be involved in relationships corresponding to many of these phenomena, sometimes all. Although there is much overlap, they have different characteristics, and to create broad change, there is a need to enter conversations within all, or most, of these contexts. On the other hand, most of the conversations exist independently of action research and do not need to be created from scratch. Even though they are many, there is also an advantage associated with their specialization. Action research does not have to take part in conversations that are only vaguely relevant.
Although with exceptions, the focus has shifted toward continuously larger systems. While the points of origin were the groups of six organizations participating in the job design seminar, and the units of similar size that constituted some of the enterprise groups in Enterprise Development 2000, much of the focus is presently on branch-based innovation systems and regional and national configurations of organizations.
With this, the role of action research is located within the framework of a set of mutually dependent conversations. But what are, more specifically, the contributions of action research within these contexts? Turning back to the function of the first experiments, they were generally thought to demonstrate the relevance of psychological job requirements and specific socio-technical design criteria (Emery & Thorsrud, 1976; Trist, 1981). While this was certainly the case, there seems, however, as if it was a third, and slightly different, perspective that came to be decisive in the long run: the potential of constructive efforts performed jointly between the workplace actors. While this point may look trivial today, there is a need to remember that the first experiments occurred at a time when the discourses on industrial society and its moving forces were quite heated between, on the one hand, the proponents of economic liberalism, and on the other those of the various branches of socialism. These world views tended to confront each other since they both claimed universal validity. Generally, the founders of Scandinavian social democracy became known for their ability to “mix” elements from the different schools. To be able to mix elements, they had, however, to reject the absolutism of both, and reduce them to sources of ideas. With this, they also came to reject the belief that the world would be set right if and when one of the views became dominant—in a more or less distant future. Instead, their point of departure was that society was open to choices and constructions: it was up to the people concerned to decide what they wanted, and to act accordingly.
This kind of constructivism called for all actors involved in a field, such as working life, to play a role in the process. The reason was not “participative idealism” but simply the point that any specific interest group would either have to be given a voice or it had to be dissolved. In this way, the pioneers of Scandinavian social democracy avoided the stalemate discussions on the roles of unions and employers characterizing societies with a greater belief in general theory. The market liberalists never liked the unions, but were, in most societies, not able to fully get rid of them. Likewise, the socialists always looked askance at the employers, but were, again, unable to get rid of them. Accepting that both, along with a number of other interest groups, had to participate, the social democrats were looking for ways in which this mutual participation could be organized.
Since what is now often called the Scandinavian model originated with a cooperation between the government and the labor market confederations, it acquired a strong centralist orientation. Work and the labor market had to be reformed, to reduce the level of conflict and increase productivity, but the platform had to be an agreement between the central actors. From this platform, a cascade of agreements, training systems, and conflict resolution mechanisms emerged until, eventually, the workplaces were reached. It was up to the politicians to see to it that the benefits of peace and productivity were divided in a fair way, through policies within areas such as employment, welfare, labor protection, and taxation. Under this umbrella, Sweden, in particular, saw a major wave of implementation of Taylorism as early as the 1930s (Johansson, 1978), which was to be followed by the other Scandinavian countries in the post World War II period. This development was, however, not free of problems. The continuously intensified implementation of Taylorism led to unrest in the workplaces, and the centrally administered health and safety procedures seemed to fall far short of the real needs. Major challenges posed by work seemed to avoid centralist constructivism. This was the context where the experiments with autonomy in work appeared. While autonomy and associated notions of psychological job requirements and socio-technical design attracted interest, it seems clear that it was the local constructivism as such that attracted most of the attention: The ability of managers, workers, unions, and other stakeholders to get together locally and through local processes find ways of meeting the major challenges of working life.
But how did the experiments function? At the core of the experiments was the idea of taking abstract concepts—like democracy, participation and autonomy—out of the general conversations of society and confront them with workplace realities. By workplace realities was, however, not understood “the workplaces as they are”, but the workplaces as they are when the local actors confront the task of promoting autonomy, which is a rather different perspective and the one that sets action research apart from pure descriptive-analytic research. Through this confrontation with workplace realities such concepts as democracy, participation, and autonomy became charged with new meanings associated with the acts needed to make the concepts into workplace realities. This perspective can be applied to later efforts and projects as well: for local experience to exert influence on society level concepts, there is a need for a local as well as a society level discourse and there is a need for discourses that link them to each other. Participation in the health and safety reforms as well as in the evolution of the agreement between the labor market parties on workplace development was actually directed mainly at the context of workplace change. As far as the last is concerned, there have, in fact, been a large number of projects over the years, but they have been tailored to fit the contexts prevailing at each and every time. This is a main reason why they often lack universal visibility; they can be seen only from the standpoint of specific discourses and the challenges and concepts appearing as important within each of them.
This has implied a mix of tasks and roles for action research, ranging from designing field experiments, via the organization of discourses, to activities aiming at linking organizations to each other in continuously larger systems, and further on to contributions to the institutional order of society in terms of agreements and legislation. “The optimal way” of performing each of these tasks varies with context. Action research will not improve on its performance through improving on one, or even some, of these activities. The main focus even in contemporary action research on how to make better individual cases is only one of a number of challenges. In fact, a focus on optimizing on single case level may be counterproductive. As demonstrated by Kania and Kramer (2011), as soon as a case is part of a broader network of cases, the characteristics of this broader network decide what is optimal for each case.
From general theory to ripples in the water
While the early field experiments in Norway represented outstanding cases, there is a notable lack of them in contemporary Scandinavia. Events as early as the job design seminar indicated that organizations could move forward through working together and using experience from everyday rather than outstanding situations. At the same time, attention shifted from single organizations toward the relationships between organizations. An important experience was that within each set of relationships between organizations, there was a need for a pattern of give and take; all needed to learn something but all could function as a source of learning for others. Insofar as something could stand out, it would have to be the configuration of organizations rather than the single organization. However, as the number of networks started to grow, and networks were brought to co-operate within regions, the perspective of equal partners rather than leader–follower was applied to the relationship between networks as well. Ultimately, this leads to a main emphasis on lifting working life as a whole through parallel processes. The outcome is a population of organizations with the shape of a ball: there are few at the top but there are few at the bottom as well. Most cluster around the middle. The organizations can still perform well and the national economy be strong if this average is (relatively) high. So far, this seems to be the case as the Scandinavian countries are concerned. It means that in creating local projects there is a need to have a certain number, and it is more important to make many move some distance than to make a few move a major. For a case, it is less important to stand out than it is not to stand alone.
But is it possible for action research to generate not only a broad range of cases, but to perform a continuous renewal of these cases as well? The point that the development sketched above spans half a century indicates that there are good possibilities for spreading activities in time. Broadly framed programs, involving a number of different action research groups, have been applied to mobilize resources on a certain scale. How to do this? Is it not, to make researchers work together, necessary to make all subscribe to the same theory and the same procedures; an impossibility? What is given in the programs are, however, not full specifications for how the local projects are to be performed, but a framework that the local projects need to stay within. The programs have left room for very different schools of thought concerning organizational change, ranging from highly structuralist perspectives to radical linguistic ones. In Scandinavia, all the programs have attracted more researchers and users than what the programs have been able to support.
Since the point about having a number of projects is to be able to operate under a broad range of different local conditions, it follows that there is no “best way” in which to promote autonomy in work on the local level. The efforts performed by the different research groups do not represent a joint implementation of one single theory of organizational change, but rather the expressions of different approaches in what can probably best be called a social movement. A social movement is characterized by shared goals that are pursued along different paths but in such a way that participants learn from each other (Gustavsen, Hansson, & Qvale, 2008). Conditions that pertain in the same way to all are those that are expressed in institutional conditions on the level of society. Although various forms of research contributions have helped shape the frameworks, the frameworks must not be taken as “theory” subject to replacement by “better theory.” They may rather be seen as contracts between the major interest groups in society expressing common intentions and procedures. They are subject to revision, but in the light of considerations that span well beyond those that can be settled by research arguments alone.
The role of action research must be seen in terms of a movement where researchers from different schools work together to promote participative constructivism in the form of the ability of those concerned to themselves create and shape the society in which they live. It is these broadly framed movements that represent the primary generalization of action research knowledge, not textually expressed claims to generality, nor the specific measures applied to broaden and strengthen the movements.
This perspective can be applied to a number of the developments in which action research has been, and is, a part. In this article the movement towards autonomy and democracy in work has been used. Other examples are the movement toward equal treatment of women and men in work and society, towards self-reliance in the third world, and to save the planet from ecological disaster. In all cases action research has been—and is—more or less strongly involved and has exerted at least some influence over the overall nature of the movement. The movements differ in terms of scope, membership, types of ideas, forms of relationships, and more, so do the contributions of action research. It seems, however, as if the organization of confrontations between the general and the local, the abstract and the concrete, constitutes a common element. What most action researchers do is to conduct one or some projects, and use these projects to back specific lines of argument in conversations on various levels in society. For research contributions to promote and sustain social movements, there will generally be a need for contributions from more than one researcher or research group. However brilliant a small group may be, it will not be able to organize changes of the scope and magnitude needed to sustain a movement. There is, consequently, a need for some degree of cooperation, or at least synchronization, among a number of researchers.
As an overall characteristic of this kind of development the notion of social movement is used. There are challenges associated with this choice of concept. First, that it appears as a highest level in a hierarchy of levels of organization, spanning from networks and clusters via regions and innovation systems to nation states and corresponding organizations, such as the labor market parties. On each of these levels, or systems, there is a broad literature, and the characteristics of each type of system are generally cast in different terms and with little concern for the characteristics of the other levels. Second, as the notion of social movement appears in the literature—largely political science—the emphasis is generally on leadership and the conditions that create the movements, less on the grass-root actors as subjects (McAdam, McCarthy, & Zald, 1996). The closest alternative may be “community,” but this seems to be too narrow. Although Norway has been used as an example in this contribution, the action research inspired notion of autonomy has, although through somewhat different channels, diffused to substantial parts of the total labor market in Scandinavia, a labor market of about 15 million people. The purpose of this contribution is, however, to draw attention to the issue of the wider impact of action research and—hopefully—to inspire more action researchers to explore this issue, not to freeze a specific set of concepts. If and when more experience is documented and discussed, one may come to choose other concepts as more fruitful.
The arguments promoted in this article do not call for major changes in the action research establishment. What they do call for are more attention given to the issue of how to reach out in society, greater willingness among action researchers to co-operate with each other, and, perhaps most important, a major improvement in the ability to tell society what action research actually achieves in terms not only of change but in terms of the scope of the change. In spite of the point that action research may, in general, be on the right track and can be found in a large number of countries, it remains, after three quarters of a century, a marginal part of the research establishment. There is no universal theory that will enable action research to break out of this encapsulation. It can be done only by utilizing its major asset to the full: creating change in the real world. Change is, however, not enough. The world must be told what change is achieved and there must be proof behind what is told.
