Diversity and constructive conflict in stakeholder dialogue: considerations for design and methods

The nature of wicked problems suggests that problem boundaries are not given but need to be probed. Different stakeholders have different perspectives on a problem and its potential solutions and accordingly different problem definitions. This emphasizes the importance of stakeholder dialogue for problem definition rather than problem solving. There have been several contributions to the policy sciences field that highlight the pitfall of trying to solve a problem without exactly knowing what the problem is (e.g. Cockerill et al. 2009; Hisschemöller and Hoppe 2001; Van de Kerkhof 2006a, b). This pitfall has been referred to as ‘type III error’, which means ‘solving the wrong problem’ (Mitroff 1974; Raiffa 1970: p264; cited in: Dunn 2001).¹ A strong focus on problem solving wrongfully suggests that participants are well aware of others and their own position. In the case of wicked problems, especially, this is often not the case. Reflecting the argumentative turn in policy analysis and planning (Fischer and Forester 1993), a focus on problem definition rather than problem solving means that a stakeholder dialogue is a problem structuring activity: “socio-political interaction aimed at becoming aware of a problem through generating, using, exchanging, confronting, evaluating and integrating as much (contradictory) information as possible, which is enclosed in causal, normative and means-end presumptions about the problem and its solution”(Hisschemöller 1993).

The aim of stakeholder dialogue as a problem structuring is thus to learn about a problem and its solutions (see Dunn 2004). More specifically, learning in stakeholder dialogues on wicked issues means gaining an improved understanding of the diversity of perspectives on the problem and its potential solutions (see also Renn 1999). Diversity of perspectives is critical for enhancing learning processes on complex issues (Schweiger et al. 1986; Jehn et al. 1999 who also showed that this is not the case for routine tasks). Learning in this sense starts with realizing that not only one unique perspective exists, but at least two (Müller-Merbach 2004). A perspective is defined as the integrated whole of beliefs, values and presumptions that people use to get to grips with a particular problem (similar to a frame: Rein 1986; Schön and Rein 1994; and to a policy theory: Hoogerwerf 1990). A perspective shapes a person’s perceptions and determines how someone perceives a particular problem and its solution. It represents a way of making sense of and acting upon reality.

Yet diversity of perspectives is no guarantee for learning. All kinds of mechanisms exist that hamper an open exploration of divergent perspectives (Cuppen 2009; Stasser and Titus 1985; Greitemeyer and Schulz-Hardt 2003). For instance, people tend to take up information that underlines their initial ideas, rather than information that conflicts with their initial ideas. Also, the influence of jargon may lead to situations in which some people fully understand what is being discussed and others not at all (this is for instance even the case when different scientific disciplines interact). This paper suggests constructive conflict as a mechanism to enhance learning in stakeholder dialogue. Constructive conflict refers to an open exploration and evaluation of competing ideas and knowledge claims in order to achieve new ideas, insights and options for problem solving. It takes place through a process in which participants confront each other’s claims with their own claims, unravel argumentations, make (implicit) assumptions explicit, and jointly develop new ideas that are more robust. The term has been introduced before by Amason et al. (1995), and it has also been referred to as productive conflict (De Dreu 1997; Jehn 1997) and creative conflict (Müller-Merbach 2004). Although the word ‘conflict’ may have negative connotations, diversity of perspectives is actually the reason why stakeholder dialogues are thought to be effective in the first place (Amason et al. 1995). Critical for stakeholder dialogue is the question how conflict remains constructive rather than destructive. This question will be addressed in the next section.

A stakeholder dialogue enhances learning through deliberation rather than consensus building (see Van de Kerkhof 2006a, b). This does not mean that a stakeholder dialogue cannot result in consensus. Rather, it means that participants are not forced to reach a consensus, as this may put impediments to the creation of useful insights for policy-makers and stakeholders and may lead to the adoption of invalid assumptions and/or inferior (policy) choices. Mitroff and Emshoff (1979: p10) state that: “the danger is not in reaching compromise, but in reaching it too soon and for the wrong reasons, e.g. because of the inability to tolerate conflict as a sometimes necessary and valuable tool for policy making”. Instead of focusing on consensus, a focus on confrontation of competing viewpoints prevents shifting away from reaching a quality decision towards reaching an agreeable one (Coglianese 1999 cited in Van de Kerkhof 2006a, b). A stakeholder dialogue is thus open-ended, in the sense that it does not necessarily result in one consensus policy option or strategy. With learning as the aim, closure in the problem structuring process ideally results in a synthesis of divergent perspectives. This means a new, robust understanding of the problem and its potential solutions. Synthesis contains an assessment of the problem and divergent ways for dealing with the problems, with the broad range of perspectives as input. So, rather than resulting in one consensus or compromise outcome, it results in an overview of divergent ways for dealing with the problem, taking into account congruency (Grin and Van de Graaf 1996) between perspectives.

The fact, that boundaries of wicked problems are not given but need to be probed, suggests that the design of stakeholder dialogue needs to avoid imposing predefined boundaries. Dunn (1997) refers to this as ‘estimating the boundaries of ignorance’. He uses the notion of ‘congruency’ to indicate that methods to support problem structuring should fit the type of problem under investigation. Congruency is especially important with relation to stakeholder selection procedures. A congruent method for stakeholder selection is based on the empirical identification of perspectives. Random sampling for example is not congruent, as it is based on invalid problem perceptions, such as an even distribution of relevant perspectives among the stakeholder population. Also, the often-used selection procedure that is based on actor type (invite stakeholders from industry, academia, NGOs, policymakers, etc.) is not necessarily congruent as it is based on the assumption that diverse stakeholder types reflect diverse perspectives. This is, however, an empirical question (the assumption was shown to be incorrect for the case of sustainable biomass: Cuppen et al. 2010). The design of stakeholder dialogue should thus foster an open and unbiased process without (implicit) assumptions on questions such as who can deliver facts to the process, who can deliver values to the process, which variables need to be included, which facts are (un)contested, etc. Rather than assumed, these questions need to be empirically investigated. This reflects the constructivist notion that facts and values are strongly interwoven for (policy) problems. What is a fact to one person may be a value to another person and as a consequence conflicts can arise through conflicting values as well as through conflicting knowledge claims (Hisschemöller 2005). This paper will investigate how and which (participatory) methods can support such an open and unbiased process of problem definition.

The constructivist notion implies that methods that are based on a distinction between facts and values are not helpful for structuring wicked problems in stakeholder dialogue. To clarify this claim, we can look at some public participation methods that implicitly make this distinction. Methods such as consensus conference (Joss 2000; Einsiedel and Eastlick 2000; Oughton and Strand 2004), planning cell (Dienel 1978), cooperative discourse (Renn 1999, 2004) and citizen’s jury (Crosby 2003; Huitema et al. 2007) are based on the idea that participatory methods are a way to combine technical expertise and rational decision-making with public values and preferences (see e.g. Renn et al. 1993). This presumption reflects a distinction between experts and laypeople and attributes ‘technical expertise’ to the domain of experts, and ‘values and preferences’ to the domain of laypeople. The distinction between experts and laypeople seems to reflect a perception of certainty on the relevant expertise for dealing with the problem, although no such certainty may really exist. Moreover, as many differences as there may be between experts and laypeople, just as many differences may there be among experts and among laypeople. Rather that presuming which groups can contribute in what ways, the question ‘who has relevant expertise and would be willing to contribute?’ is more interesting. In that sense, Collins and Evan’s suggestion to abandon the oxymoron ‘lay expertise’ is quite appealing (Collins and Evans 2002).² They argue that—as the dictionary definition of ‘layman’ includes the meaning ‘someone who is not an expert’—it makes more sense to talk about ‘experts’ than ‘lay experts’. Lay people are experts with respect to their own problems (Mitroff et al. 1983), although the expertise of these (former lay-) experts has not been recognized by certification (Collins and Evans 2002). Expertise is not restricted to technical or scientific expertise, but it can be experience-based or related to particular values or interests. The notion that laypeople have particular expertise to offer will probably not be denied by the scholars working on the methods mentioned above. The point is, however, that stakeholder dialogues on wicked problems should go beyond the lay-expert distinction. Method and design for stakeholder dialogue should treat the identification of stakeholders with relevant expertise (regardless of whether he is an ‘expert’ or a ‘layperson with expertise’) as an empirical question rather than a (implicit) assumption.

This question can be tackled by using congruent methods for stakeholder selection. That is, if we say that stakeholders have particular expertise because of their perspective, we need to find methods that can empirically investigate perspectives and select stakeholders on the basis of those perspectives. This approach allows for involving different types of expertise without defining them beforehand.

The idea that diversity is an important merit of stakeholder dialogue is obviously not new. However, what is meant with diversity often lacks definition. In order to understand how exactly stakeholder dialogue can benefit from specific forms of diversity, three properties of diversity need to be distinguished. These three properties are the following (Stirling 1998): variety, balance and disparity. ‘Variety’ refers to the number of categories into which the elements can be divided. ‘Balance’ refers to how the elements are distributed among the categories. ‘Disparity’ refers to the degree and nature to which the categories themselves are different from each other. As an illustration of these three properties of diversity, let us suppose we have to evaluate a policy for the diversification of neighbourhoods. One thing we could do is assess the diversity of neighbourhoods in terms of nationalities. The three properties of diversity can be used for operationalization of diversity; in other words, when is a neighbourhood diverse in terms of nationalities? The bigger the variety of nationalities is, the bigger is the diversity of the neighbourhood. But the neighbourhood will be also more diversified when the number of people per nationality is balanced. For instance, a Dutch neighbourhood in which many Dutch people live and only a few Turkish people is not very diversified. Finally, a neighbourhood will also be more diversified if there is a certain extent of disparity between the nationalities. For instance, a Dutch neighbourhood with only Western nationalities is less diversified than a Dutch neighbourhood in which also Asian nationalities live. Disparity is a qualitative dimension of diversity. So, whereas anyone would probably agree that a Dutch and a German person are more similar in terms of nationality than a Dutch and a Chinese person, disparity will often involve a (personal) evaluative judgment. However, the same can be said for variety, as the categorization of elements is not fixed. It will depend on someone’s particular perspective how variety and disparity are constituted.

From research on working groups, it appears that variety can improve the quality of group decision-making: heterogeneous groups produce higher-quality outcomes (better strategies) than homogeneous groups (Hoffman 1959; Hoffman and Maier 1961). Although the working groups in these two studies were heterogeneous with regard to personality (measured through a standard survey to assess personality traits) rather than perspective on problem and solutions, this effect also seems to hold for groups that are heterogeneous with regard to perspectives. The inclusion of a variety of perspectives induces more divergent thinking, consideration of multiple perspectives and consideration of higher proportions of unshared information (Brodbeck et al. 2002). According to Dunn (2001), marginal viewpoints, or rarely mentioned hypotheses, have more probative value than hypotheses mentioned more frequently—those on which there is substantial consensus. Highly probable or predictable hypotheses do not challenge accepted knowledge claims. In other words, one will probably learn more from something never heard before than from something already familiar with.

So, not only is it important that a variety of perspectives is included in a stakeholder dialogue, but also the inclusion of marginal perspectives is critical for learning. This refers to the disparity property of diversity: the more different an idea is, the larger its learning effect will be. This idea is supported by research by Brodbeck et al. (2002) who state that “minority influence facilitates open-mindedness towards alternative solutions”. They showed that the likelihood that a group identifies new and qualitatively superior decisional alternatives increases as a result of “minority-induced divergent thinking”. This idea can also be found in innovation systems literature and institutional analyses, in which it is argued that it is important to involve, e.g. small entrepreneurs because incumbent system players are usually not the first to initiate successful system innovations (Agterbosch 2006; Breukers and Wolsink 2007; Hekkert et al. 2007). Marginality refers to the newness of perspectives for stakeholders: a marginal perspective is a perspective that is not often heard in the dominant debate about the issue under study. This does not necessarily mean that there are only a small number of stakeholders adhering to the perspective, neither does it mean that a perspective that only a small number of stakeholders adhere to cannot be dominant.

In addition, the balance property of diversity is important. It appears that groups in which variety is balanced are more likely to disseminate unshared information than unbalanced groups (Brodbeck et al. 2002). Furthermore, it has been argued that the balanced inclusion of perspectives reduces groupthink (Janis 1972; Dryzek and Niemeyer 2008). This suggests that learning in stakeholder dialogue can be enhanced by the participation of an equal number of stakeholders from each perspective. This means that, regardless how dominant or marginal the perspective is in terms of percentage of the whole stakeholder population, every perspective is equally present in the dialogue.

Conflict is not in every form beneficial to group processes (see e.g. De Dreu and Van de Vliert 1997; Jehn and Mannix 2001; Jehn et al. 1999; Jehn 1995; Kruglanski and Webster 1991). Three particular issues appear to be important when it comes to constructive conflict in groups. First, when it comes to learning from minority dissent, learning seems to benefit more from authentic conflict than from artificial conflict, as is the case in (some instances of) the devil’s advocate approach (Nemeth et al. 2001; Nemeth et al. 2004). Nemeth et al. (2001) investigated in their study whether groups in which someone was assigned the role of devil’s advocate and who was known to believe that position (the so-called consistent devil’s advocate condition) produce similar results as groups in which someone was not assigned a particular role in the group (the so-called authentic minority condition). In both conditions, the person advocated a dissenting minority viewpoint. Their study shows that groups in the consistent devil’s advocate condition produce fewer solutions to a specific problem than groups in the authentic minority condition. In addition, the number of quality solutions (as judged by two independent raters) is higher for groups in the authentic minority condition than for groups in the consistent devil’s advocate condition. This means that the mere fact that someone is assigned the role of devil’s advocate makes groups less productive in the sense that these groups produce fewer solutions and fewer good solutions to a specific problem than groups in which the dissenter is not role-playing. This may be very specific for minority dissent, as other role-playing and gaming literature suggest that role-playing is a powerful tool for learning (see for an overview of literature Crookall 1995). For stakeholder dialogues, this is an important finding, as it suggests that devil’s advocate methods are not most suitable for stimulating constructive conflict.

Second, conflict seems to enhance learning only when it is cognitive rather than affective (Jehn 1997; Amason et al. 1995). Cognitive conflict “pertains to conflict about ideas in the group and disagreement about the contents and issues of the task” (Jehn 1997), whereas affective conflict “exists when personal and relationship components within the group are characterized by friction, frustration and personality clashes within the group” (Jehn 1997). Affective conflict typically includes tensions, animosity and annoyance among members within a group (Jehn 1995). The distinction between cognitive and affective conflict is not based on the object of a discussion (for instance facts versus values), but on the nature of a discussion. Following the constructivist notion that it does not make sense to distinguish facts from values, conflicts can arise on both, taking the form of cognitive as well as affective conflict. It appears that in groups performing non-routine tasks (which a stakeholder dialogue is too), cognitive conflict is beneficial for task performance (Jehn 1995; Jehn 1997). Groups that debate and criticize each other’s ideas are more creative than groups that do not allow conflict (Nemeth et al. 2004). However, affective conflict appears to be detrimental for group performance (Jehn 1995; Jehn 1997). It is of course the question to what extent in practice cognitive and affective conflict can be separated. Persistent cognitive conflict may very well lead to affective conflict, including all kind of affective reactions such as annoyance and animosity (De Dreu 1997).

Third, conflict needs to be manageable, in the sense that people do not feel overwhelmed by it (Jehn 1995). It is likely that there is an optimum relation between conflict intensity and group performance (De Dreu 1997). Similarly, there is an optimum relation between “newness” and “familiarity” of information, which is referred to as ‘optimal cognitive distance’ (Nooteboom et al. 2007). The concept ‘optimal cognitive distance’ suggests that there is an inverted U-shaped relation between ‘cognitive distance’ and ‘innovation performance’: “In first instance, as cognitive distance increases, it has a positive effect on learning by interaction. When people with different knowledge and perspectives interact, they stimulate and help each other to stretch their knowledge for the purpose of bridging and connecting diverse knowledge. […], However, at a certain point cognitive distance becomes so large as to preclude sufficient mutual understanding needed to utilize those opportunities.” (Nooteboom et al. 2007). The concept ‘optimal cognitive distance’ is also important in relation to the learning potential of new or marginal perspectives. It was argued above that one would learn more from something never heard before than something already familiar with. The concept ‘optimal cognitive distance’ suggests that this is only the case when the information is not too different or new, but rather ‘sufficiently’ different or new. For example, I could have difficulties in understanding, and therefore learning from, an advanced MSc astronomy course, because this information is too new or unfamiliar to me. A way to deal appropriately with the newness of information in a dialogue might be to include people who can bridge disparate perspectives or claims and as such can serve as mediators or translators of specific information.

Stakeholder identification and selection is a critical step in the design of stakeholder dialogue, as it is decisive for the validity of the dialogue. That is, no matter how well the dialogue itself is designed, if the stakeholder selection procedure was not able to identify and include the variety of perspectives, problem structuring in the dialogue is seriously hampered. There is a plethora of methods to facilitate participatory processes (see for an overview e.g. Rowe and Frewer 2005). However, methods to facilitate stakeholder identification and selection remain rather underexposed. Much attention is paid in literature to the representativeness of participatory processes, “offering criticism of their frequent failure to reflect the characteristics of wider society in the groups used” (Martin 2008). It is, however, rarely clear which criteria are used to judge whether a selection procedure is representative (Crawford et al. 2003: p 46; cited in Martin 2008; see for a more detailed discussion of representativeness and types of sampling Cuppen et al. 2010).

Building on the three properties of diversity and their contributions to constructive conflict, representativeness is understood here as a stakeholder sample reflecting a balanced inclusion of the variety of perspectives that exists within the stakeholder population. Dryzek and Niemeyer (2008) refer to this as ‘discursive representation’. A balanced inclusion of the variety of perspectives means that the share of marginal, or disparate, perspectives in the dialogue is equal to that of other perspectives (in terms of number of participants). This implies that the distribution of perspectives in the dialogue, i.e. the sample, does not reflect the distribution in the wider stakeholder population. Importantly, it means that stakeholders are being selected because of their perspective on the issue under consideration rather than their affiliation or the fact that they belong to a particular group (this reflects the notion of congruency of selection procedures introduced in “Wicked problems and stakeholder dialogue design”). So rather than using actor categories as a guideline for stakeholder selection (e.g. stakeholders from industry, (local/national) government, NGOs, academia), perspectives are used as a guideline for stakeholder selection. This raises the need for the identification of perspectives before stakeholder selection can take place.

In the second section of this paper, it was discussed that the wicked character of the problems under study implies that problem boundaries need to be probed. It was argued there that the design of stakeholder dialogue needs to avoid imposing predefined boundaries and that it should avoid (implicit) assumptions on questions such as who owns what kind of relevant expertise. This means that bottom-up methods are needed for identifying the diversity of perspectives, which make it possible to link stakeholders to perspectives. Methods should assist the empirical identification of the classes within the diversity (i.e. perspectives), rather than assuming diversity through some kind of category system⁴ (such as stakeholder selection based on socio-demographic variables, actor types, personality traits, or value orientations such as in cultural theory: Douglas and Wildavsky 1982; Thompson et al. 1990).

In order to be able to identify the diversity of perspectives, it is important to cut across stakeholder networks. Within networks, people know each other and each other’s ideas; hence, knowledge and perspectives are likely to be more redundant than diversified (Granovetter 1973). It is therefore less likely that people from the same network will be confronted with new ideas or perspectives than when they discuss with people from other networks. This refers to the disparity property of diversity. Hence, in order to make sure that disparate perspectives are included, it is important to identify stakeholders from different networks.

Identification of perspectives and selection of stakeholders are strongly related in the design of stakeholder dialogue. Both take place in the preparation phase of the actual dialogue. Then, in the dialogue itself, learning through constructive conflict continues. In order to learn from the variety of perspectives present in the dialogue, perspectives need to be articulated. This can be done by clarifying argumentations on specific problem issues and relating these to the stakeholder perspectives (see for an example next section). Emphasis is on the implicit or taken-for-granted elements of perspectives, such as implicit assumptions. People are often unaware of their own perspectives, and their behaviour is often guided by tacit, implicit or taken-for-granted beliefs and presumptions. By making implicit elements explicit, perspectives can be unravelled, and as such learning about how different people look at reality and how they define a problem can be enhanced. This helps to understand what a particular perspective encompasses and how perspectives differ from, and relate to one another.

In the dialogue, articulation of perspectives can take place in ‘like-minded’ subgroups (in line with assumptional analysis, see Mitroff et al. 1979; Mitroff and Emshoff 1979) to stimulate deliberation. That is, average speech time in smaller groups is higher, and shy or cautious persons may be more inclined to speak in a small group than in a large group. This means that participants with similar perspectives can jointly work out argumentations for specific claims based on their particular perspective. The operationalization of perspectives depends on the specific aim, context and topic of the participatory process one is engaging in.

The knowledge claims, ideas and presumptions that resulted from the articulation of perspectives are compared and confronted in the dialogue. Prepared by the first two steps, this is the step in which constructive conflict should now fully evolve. It involves the challenging of claims, ideas and presumptions, and it should stimulate participants to engage in a process of reflection, rethinking or frame reflection (Schön and Rein 1994). The presumption is that when challenged to consider other, maybe even conflicting claims, participants will develop a better understanding of other ideas and the relation to their own ideas. This may either strengthen or change their own ideas.

In the third section of this paper, three issues were discussed that appear important for constructive conflict in stakeholder dialogue. Implications for methods and design can be derived from these issues. First, conflict appears to be more constructive when it is authentic than when it is artificial. This is ‘good news’ for stakeholder dialogue, in which ‘real’ stakeholders, with ‘real’ expertise, interests and values participate. Confrontation between these ‘real’ stakeholders and their ‘real’ perspectives enhances constructive conflict (more than e.g. role-play would do).

Second, conflict is only constructive when it is cognitive rather than affective. In terms of methods and design, cognitive conflict can be stimulated by the confrontation on a specific level. This means that the objects of discussion are specific technological or policy options, and that confrontation takes place on the level of knowledge claims rather than perspectives. Confrontation of perspectives will not result in a fruitful dialogue; there is no possibility of falsifying perspectives, as perspectives are the lenses through which people make sense of social reality (Schön and Rein 1994). Confrontation of perspectives will therefore result in conflict, but not in constructive conflict. Rather, it will create a deadlock, because no one perspective is ‘better’ or ‘more true’ than another. Confrontation of perspectives will more likely result affective conflict, which has been shown to be detrimental for group processes. It is only at the level of specific technological or policy options that people with different perspectives can ‘find’ each other (Grin and Van de Graaf 1996; Guba and Lincoln 1989). Yet the concrete level of discussion is no guarantee for cognitive conflict. As was mentioned above, there may always be a risk that cognitive conflict turns into affective conflict. This needs to be avoided and should be an important point of attention for the facilitator of the dialogue. It needs to be emphasized in the dialogue that participants are not required to reach a consensus or an agreement and that they are allowed to disagree.

Third, conflict needs to be manageable, i.e. there is an optimum in the relation between the distance between perspectives and the learning potential. So, whereas the inclusion of disparate perspectives enhances constructive conflict and thus problem structuring, it may be necessary to include stakeholders with ‘intermediate’ perspectives who can serve as ‘translators’ or ‘bridges’ between disparate perspectives. Another way to manage conflict is by dividing participants into smaller subgroups, each of which is chaired by an external chairperson. This chairperson should be someone who is perceived as being knowledgeable and independent, in the sense that he or she can openly engage in an exploration of competing viewpoints, and when necessary, can bridge disparate perspectives or ideas.

Q methodology (Stephenson 1935; Brown 1980; McKeown and Thomas 1988) was developed as a method to study subjectivity. It can be used to identify the range of perspectives on a particular issue. Q methodology—a combination of qualitative and quantitative method—has been applied in various fields of social science. It has for instance been used to identify views regarding citizenship, the public interest, environmental policy and the quality of participation processes (Clarke 2002; Dryzek and Berejikian 1993; Ellis et al. 2007; Van Eeten 2001; Webler et al. 2001; Barry and Proops 1999; Davies and Hodge 2007; Swedeen 2006; Wolsink 2004; Breukers 2006).

Three characteristics of Q methodology make it especially relevant for stakeholder dialogue. Firstly, Q methodology is open with respect to the boundaries of the problem under consideration. Q methodology empirically constructs perspectives without imposing predefined categories. The merit of Q methodology is that “by allowing the categories of the analysis to be manipulated by respondents, the researcher loses the exclusive power to signify the reality of the researched” (Robbins and Krueger 2000, p. 645: 645). This in contrast to for example surveys or questionnaires which include specific items supposed to measure specific pre-defined variables. Q methodology identifies respondents’ views on statements in relation to all statements presented (see e.g. Dryzek et al. 1993), whereas in surveys and questionnaires respondents are asked to express views on isolated statements. Secondly, Q methodology allows for diversity in stakeholder perspectives. It enables addressing the three properties of diversity: variety, balance and disparity. Q Methodology traditionally aims to give a picture of the variety of perspectives among the population, rather than analysing the level of support for those perspectives among the population (the balance of perspectives) as in traditional surveys and questionnaires. This implies that the procedure for sampling respondents is usually different from that in surveys and questionnaires. Rather than random sampling and large sample sizes, Q methodology relies on purposive sampling and smaller sample sizes. The fact that there is a person who can be assumed to have a different point of view is enough reason to include him or her in the sample. Thirdly, Q methodology allows for statistically linking stakeholders to perspectives. This characteristic makes it possible to identify for example the stakeholders who most strongly represent a particular perspective, or who are a ‘mix’ of multiple perspectives. As a consequence, Q methodology is very useful as an empirical method to select stakeholders for participation in a dialogue. It makes it possible to select participants on the basis of perspectives rather than for example affiliation.

The central task in Q methodology is the Q sort. Each respondent ranks a set of statements on a (usually normally distributed) scale that represents significance or salience (Brown 1980 : p198), such as most agree to most disagree. Statements are selected in such a way as to reflect the diversity of the ‘concourse’. The concourse is the full range of discussions and discourses on the particular issue under study (identified through all kinds of written and/or spoken sources). Factor analysis of the Q sorts results in clusters of Q sorts (factors) that can be interpreted as perspectives. The Q sorting task is often accompanied by an interview, in which respondents are asked to explain their sort—this helps to interpret the factors (see for a detailed description of the procedural steps: Davies and Hodge 2007).

Q methodology was employed in the dialogue with a threefold aim. First, the aim was to empirically construct the variety of stakeholder perspectives on sustainable biomass in a bottom-up way. Second, the aim was to select stakeholders for participation in the dialogue accordingly. Stakeholder selection was supposed to result in the balanced inclusion of the variety of perspectives, including disparate or marginal perspectives (a detailed account of this application of Q methodology can be found in: Cuppen et al. 2010). Third, the aim was to structure the dialogue on the basis of the perspectives.

To cover the broad range of perspectives, 200 statements were gathered from reports, websites, news items and transcripts from other stakeholder projects on biomass. This was relatively easy, as the issue of biomass was heavily discussed in the media and at several public debates at the time of this study. As a consequence, the range of ideas and opinions was relatively well articulated. The transcripts from a stakeholder dialogue organized by the same project leader were furthermore very useful for identifying marginal perspectives, as it included people and ideas that were not all very well represented in the dominant debate on biomass.⁷ The set of 200 statements needed to be reduced to a more manageable size, while maintaining diversity. To validate the diversity within the reduced set of statements, three things were done. First of all, the set of statements was individually categorized by three members of the project team who then identified unique statements within categories. The three categorizations and sets of unique statements were compared and discussed, and iteratively, this process continued until sixty-two statements remained. Second, the set of statements was piloted with five people (who were all well acquainted with the biomass debate) to check whether any relevant statements were missing. This resulted in a definite set of sixty statements. Third, during the Q interview, the interviewer asked the respondent whether he or she thought statements were missing. This was in general not the case. Of the few respondents who chose to add statements, it concerned emphasizing an issue that was already covered by other statements, for instance by rephrasing a statement into its positive or negative counterpart.

The goal of respondent selection was to reflect the broad range of perspectives. Criteria for selection were that the respondent was a stakeholder who was somehow professionally involved in the biomass issue and who could potentially bring new issues or points of view to the process (which can be found in statements made in interviews, on websites, in reports or in earlier projects). Respondents were identified through newspaper articles, news–websites, earlier contacts of the project team (e.g. via the earlier stakeholder project mentioned above) and snowball-sampling technique (“please mention someone with a different, and someone with a similar perspective on energy from biomass”). Identification of respondents with diverse perspectives was also relatively easy due to the broad attention to the biomass issue. Eventually, seventy-five stakeholders performed as respondents in the Q sorting task. These respondents represented different sectors and organizations: knowledge institutes and academia, (energy) companies, branch/sector organizations (e.g. sector organization for oil and fat), small-/medium-sized enterprises (including energy consultants, e.g. working on cultivation or treatment of biomass, or on energy/heat/fuel production), NGOs, national government, regional and local government.

Respondents ranked the sixty statements according to a normally distributed 11-point scale, ranging from ‘most in line with my opinion’ to ‘least in line with my opinion’. Factor analysis of the seventy-five Q sorts resulted in six factors that were interpreted based on the statements and the interviews as six distinct perspectives on sustainable biomass (see Table 1). These six perspectives are sufficiently different in order to reflect the range of perspectives on sustainable biomass. Yet there are also issues of overlap between perspectives. For instance, both perspectives 4 and 5 see biomass as a commodity in a market in which it will eventually compete with fossil fuels, but on the condition that biomass applications have a positive energy balance. Perspective 4 is, however, more positive about the question as to whether this will be feasible and sees the solution in 2nd generation, certified biomass. Perspective 5 doubts the feasibility of a positive energy balance, as well as the potential availability of biomass. Perspective 5 is, however, not as critical as perspective 2, according to which all developments need to be stopped, as we cannot prevent negative impacts for developing countries. Perspectives 3 and 6 are both entrepreneurial perspectives. Yet perspective 6 is more pragmatic and less idealistic than perspective 3, according to which the focus should be on small-scale, decentralized applications in the Netherlands. Perspective 6 does not want to make a choice for a specific scale and type of application. Another difference between the two perspectives is that perspective 3 is very critical about the role of the Dutch government and policy.

In line with the methodological implications discussed in the previous section, the aim of this step was to identify forty out of the seventy-five respondents for participation in the dialogue, who reflect the variety of perspectives in a balanced way. The number of forty was chosen for practical reasons. Thirty stakeholders were seen as a manageable number in the context of the Biomass Dialogue and its resources; taking into account that some stakeholders would not be willing or able to participate, forty were invited. The forty stakeholders were selected on the basis of the results of the factor analysis. Factor analysis shows for each factor (perspective) a loading for each respondent. The higher a loading, and the lower the loadings on the other factors, the more a respondent’s perspective is in line with the perspective represented by that factor. This makes it possible to identify for each perspective, the respondents that most strongly represent that perspective. For each perspective, an equal number of respondents loading most strongly on the particular perspective was invited to take part in the dialogue (balanced inclusion of the variety of perspectives). This means that the more marginal and the more mainstream perspectives on sustainable biomass are included on an equal footing in the dialogue. Perspectives 1, 2, 4 and 5 could very well be recognized in the public debate on biomass at that time in the Netherlands; this was not so much the case for perspectives 3 and 6, which can be considered marginal for the Netherlands at that time. Factor analysis also shows disparity of the factors. It results in a statistical account of similarity and difference between perspectives.

Overall, about 30 people participated in the dialogue. Not all participants visited all three workshops: fifteen participants attended the first workshop, seventeen participants the second and eleven participants were present at the third workshop.

It was argued in the previous section that stakeholder selection should not only include a diversity of perspectives, but also cut across networks in order to increase the probability that people meet new people with new ideas. To check whether the stakeholder selection procedure for the Biomass Dialogue succeeded in doing this, participants were asked in an evaluation form after workshop 1 whether they had met people at the workshop, with whom they had never discussed about the biomass issue before. All participants affirmed to this question. They were furthermore asked with the ideas of whom of the other participants they were already familiar. There was one participant (an entrepreneur in pure vegetable oil) with whose ideas eight of twelve participants⁹ were familiar. In general, the picture was very mixed, and most participants were familiar with some of the other participants and their ideas and unfamiliar with some of others. Hence, stakeholder selection in the Biomass Dialogue appears to have been successful in cutting across networks.

In the dialogue, further articulation of the perspectives took place. The project team wrote a report documenting the results of the Q analysis. This report contained an extensive description of the six perspectives. In workshop 1, this report was presented, as well as the analysis of participant’s positions with regard to the six perspectives. In general, participants recognized the perspectives. They indicated that the presentation of perspectives clarified the complexity of opinions, claims and arguments. It helped them to better understand other people’s statements and ideas. They indicated, in the evaluation after the workshops, that the perspectives contributed to a constructive and open-dialogue, because the diversity of perspectives was made explicit and more understandable. Most people also recognized their ‘analysed’ perspective, but some also indicated that they felt being a mix of two perspectives. The persons, who indicated so, were in general indeed statistically loading on more than one perspective according to the Q analysis.

Based on the Q perspectives, homogeneous subgroups were formed to further articulate the perspectives by evaluating the sustainability of a series of specific biomass chains in workshop 1. The idea behind the formation of ‘like-minded’ subgroups was that, by means of deliberation, argumentations for specific ideas could be clarified and worked out, without engaging in discussions with different-minded people (see also Mitroff et al. 1979; Mitroff and Emshoff 1979). Argumentations and assumptions were made explicit by using an argumentation model as a support tool (added by the use of ‘repertory grid technique’, see Cuppen, 2009). The specific chains were selected in order to cover a variety of existing biomass chains, differing in terms of scale, biomass source, technology, end product and organization of the chain. The chains were presented with help of cartoons visualizing the chains and a short text of one A4 page at most to explain the chain.

It was discussed in the previous section that conflict is more likely to be constructive when it is authentic rather than artificial, cognitive rather than affective, and when it is manageable (i.e. based on not ‘too divergent’ perspectives). The inclusion of ‘real’ stakeholders and their ‘real’ perspectives ensured authentic rather than artificial conflict. Cognitive conflict, rather than affective, was enhanced through the use of specific existing (workshop 1) and innovative (workshop 2) biomass chains that served as a tool to derive claims and ideas about the sustainability of biomass (workshop 1) and a desirable biomass vision (workshop 2). It was furthermore stressed by the facilitators of the dialogue and the invited chairmen that participants were allowed to disagree, which contributed to an atmosphere in which participants felt open to explore divergent, even conflicting, points of view. External persons were invited for chairing the subgroups. These persons were invited by the project team because they were thought to be perceived as knowledgeable and independent, in the sense that they could engage openly in an exploration of competing viewpoints. Importantly, they were thought to be able to bridge disparate perspectives or ideas when necessary; for example to serve as a mediator between stakeholders with different perspectives or conflicting ideas. This was supposed to contribute to the manageability of conflict.

Each of the three workshops of the Biomass Dialogue resulted in a specific outcome. After each workshop, the project team produced a report describing the outcome of the workshop, which was sent to all participants. Each workshop started with a discussion of the report, and participants were asked to reflect on the report and to indicate whether they agreed or wanted to make amendments. The outcome of the first workshop was a set of criteria for judging the sustainability of biomass chains. All participants in the workshop indicated agreement on these criteria, despite the diversity of perspectives and despite the finding that different reference situations were used to judge sustainability. The second workshop resulted in three partial visions of the future about the role of biomass in a sustainable energy supply in the Netherlands. These three partial visions reflected the different ideas about the desirability of specific types of chains among participants. The third workshop resulted in scenarios describing the pathway to each of the partial visions. These three outcomes of the dialogue can be seen as a synthesis of divergent perspectives. As was noted earlier, participants were allowed to disagree, and there was no focus on consensus building. Yet the first outcome of the dialogue was a consensus outcome: agreement on a list of criteria to judge the sustainability of biomass chains. The other two outcomes, the visions and the pathways to those visions, showed divergence between stakeholders’ desired future situations and between the strategies and actions that were considered necessary and desirable. In working out the strategies and actions, it was tried to identify congruency of perspectives and visions: can we identify specific strategies and actions that fit with more than one perspective? The dialogue, however, was not able to reach this level of specificity in the time that was given for the project.

After each workshop, participants filled in an evaluation form asking for example what they learned and which parts of the workshop they did (not) appreciate. From our observations and participant responses, it became clear in the dialogue that the process of articulation and confrontation stimulated participants to reflect on their initial ideas and assumptions. Articulation and confrontation for instance resulted in the finding in workshop 1 that people arrived at different sustainability judgments because they used different reference situations. For different aspects of the chain, participants looked for different reference situations, such as an existing fossil alternative, a renewable alternative, an alternative usage of the biomass or a future situation. As such, confrontation further clarified the divergent perspectives and gave a better understanding of why people arrive at different conclusions with regard to the sustainability of biomass chains. Furthermore, the extent to which participants in the dialogue learned about the diversity of perspectives was evaluated in a quasi-experimental design (with ex post Q sorts and a control group), added by qualitative data from the evaluation forms filled in after each workshop and observations made during the dialogue (see Cuppen, forthcoming). The evaluation showed that participants in general increased their understanding of the diversity of perspectives. The perspectives were found to be helpful for participants to get grip on the complexity of the biomass issue and to understand how people with different perspectives arrive at different conclusions about biomass chains.