COMPRAM Method

7 A METHOD FOR ANALYZING AND DEFINING COMPLEX INTERDISCIPLINARY SOCIETAL PROBLEMS

In chapters two and three we discussed problem handling in general. We indicated that for handling complex interdisciplinary societal problems, knowledge of different disciplines is necessary. Therefore, complex interdisciplinary societal problems should be handled in co-operation[1]. Another reason for handling complex societal problems co-operatively (in a democracy) is that most complex societal problems have a great impact on society at large, and therfore a group widens the basis for acceptance of the interventions. As we have seen in chapter two, current knowledge of problem handling is mostly derived from the way one person solves a small domain related problem. Handling these kinds of problems differ, on some crucial points, from co-operative handling of complex interdisciplinary societal problems. The results from research on handling small, domain related problems can only be used to a certain limited extent. In handling domain related problems mostly only domain related problem solving techniques and domain related methods of analysis are used. For handling complex interdisciplinary societal problems a special method is needed that takes into account the complexity, the interdisciplinarity and the dynamic changes of the problem and its environment and that will support a co-operative way of problem handling.

As stated in chapter one, we assume that handling complex interdisciplinary societal problems can be improved by improving the quality of the problem handling skills of people, which, in turn, can be improved by using a fruitful method combined with the suitable tools.

In this chapter we will discuss some aspects of a method for handling a complex interdisciplinary societal problem. The discussion will be limited to the first sub-cycle of problem handling, to the analysis and the definition of complex interdisciplinary societal problems. We focus especially on the support of the co-operative problem handling process and discuss some of the possibilities of computer aids to this process. With this we come to expectation four:

for analyzing and defining complex interdisciplinary societal problems a method is needed that will support co-operative problem handling

what are the special aspects that a method for supporting the process of analyzing and defining complex interdisciplinary societal problems should possess?

We will begin by connecting the problem handling phases mentioned in chapter three with examples of complex interdisciplinary societal problems.

In chapter three, we indicated that there has to be an awareness of a problem before one can start handling the problem. This may seem trivial, but it may not be so obvious as it looks at first view. Some situations may have existed for a long time before they are recognized as a problem, at least by a sufficiently large group of people. Many complex interdisciplinary societal problems are not, at least not in the beginning, presented as problems. At different times and in different circumstances the evaluation of a problem can also differ. How to evaluate or judge a situation depends on time and circumstances, such as the moment in history, the political situation and the society. It also depends on the person who observes the situation, his or her societal position, the knowledge a person has and his or her theoretical concepts[2].

Although being unaware of a problem is not always serious, there are situations where noticing the problem at an early state can prevent the problem from growing too severe (Rosenthal, 1984). The concept of problem finding, discussed in chapter three, refers to this phenomenon: finding the problem in time before it becomes too difficult, too expensive or too complicated to be handled.

Personal awareness of a societal problem is not enough. In order to do something about a societal problem, or in order to stimulate others to handle the problem, there has to be agreement that there is a problem, combined with the conviction that something can and should be done about the problem.

In medical literature on somatic diseases there is a distinction between primary, secondary and tertiary prevention (Ormel, 1987; McGrath, 1984). Primary prevention means the prevention before the disease or infection takes place. Primary prevention also refers to removing the causes (Ormel, 1987). Examples of primary prevention in the medical domain include preventing a person from getting measles by giving a vaccination against it, or regularly brushing one's teeth in order to prevent tooth decay. Secondary prevention is preventing the problem from becoming worse. In the medical case, secondary prevention is prevention after the disease has started. Early location of risk-groups and risk-factors also belongs to the area of secondary prevention. An example of secondary prevention is curing a hepatitis B infection by giving the right medicines. Tertiary prevention is trying to diminish the effects of the disease, for instance giving paracetamol in cases of influenza[3].

If we use the medical categories of prevention to categorize complex interdisciplinary societal problems, an example of primary prevention might be building dikes to prevent the river from overflowing (Schaap, 1983). An example of secondary prevention is to heighten the dikes with, for instance, sacks of sand at moments of flooding. An example of tertiary prevention is evacuating people from endangered and already flooded areas[4]. Or take as another example the manure problem in the Netherlands. Until 1985 all the interventions focused on tertiary prevention only, but now the policy is to focus more on primary prevention through decreasing the livestock population (Glasbergen, 1992).

Depending on the moment when one becomes aware of the problem and depending on the problem, primary, secondary or tertiary prevention is possible.

Many complex interdisciplinary societal problems cannot be foreseen. They suddenly become apparent, unexpectedly, or they are unnoticed until they have already caused much damage. For things that can be foreseen it is usually advisable to take measures in time. But policy makers are not always willing to take the right preventative measures, for reasons of personal benefit, just plain indifference, a lack of view or it is not always possible due to lack of political or financial support.

Only problems that can be foreseen in time can be prevented. Some problems can be foreseen because they are known from past experiences. For these kinds of problems we can, if this is possible, take precautions. This kind of primary prevention is implemented on a large scale in societal and religious rules and in cultural customs. Societal rules include for instance, health care directions such as giving medical and pedagogical advice to young parents (DeTombe, 1976), legal regulations and traffic regulations. Religious rules include, for instance, rules for food preparation[5] and human behavior[6]. Cultural rules include such rules as washing hands before dinner.

Much of the policy toward 'new' complex interdisciplinary societal problems can be categorized under tertiary prevention. They serve only to reduce the negative effects of a problem. However, the efforts would be more effective if it were possible to remove the causes, to undertake primary prevention. Yet, no matter what moment it is, the previous stages of prevention should also be included in the problem handling process[7]. To really handle the problem fruitfully one should try to focus mainly on primary and secondary prevention. Only in cases where this approach is not possible should the policy makers turn to attention to tertiary prevention.

When people are aware that there is a problem that should be handled, the next question is: "whose problem is it?". This question can refer to two different matters. It can mean to "Who is confronted with this problem?; who is suffering from this problem?", or to "Who must handle it?"; "who is responsible for handling it?". As explained earlier in chapter two, the people that are confronted with the problem may not be those who are going to handle the problem. When the organization(s) who is/are supposed to handle the problem, is/are unwilling to take action, two things can happen: nothing will be done about the problem, or the group who is affected by the problem can organize itself and can try to persuade the organizations to take action. This can be done by forming an action group or a pressure group to influence policy-making[8]. There are different reasons for forming an action or pressure group. One can, for instance, form an action or pressure group based on the negative effects the problem on oneself or others[9] or form a group to protect one's own financial interests[10]. Examples of the last category are employers who urge the government to improve the infra-structure, or a group of (ex-)politicians who want to change governmental politics (Ornstein, 1993).

An important question is how to recognize a problem as a domain specific problem or as an interdisciplinary problem[11]. In policy-making and in scientific research a problem is often unjustly indicated as a mono-disciplinary[12] problem. In this case the problem is defined too narrowly, which excludes many phenomena that also influence the problem. The intervention one suggests will seldom lead to an optimal change. In order not to define a problem too restrictedly, one could start by approaching societal problems as interdisciplinary problems, because many societal problems are interdisciplinary (Crombag, 1984; Anderson, 1992). Only when it is obvious that the problem is a mono-disciplinary problem should one change this approach.

As explained earlier[14] because of the interdisciplinarity and the importance of the problem, complex interdisciplinary societal problems must be handled by a multi-disciplinary team. To analyze a complex interdisciplinary societal problem, requires knowledge from different domains. These knowledge fields are too large to be overseen by one person[15] alone. Therefore, experts of different knowledge fields should, together, handle the problem in co-operation. Each expert has a good and often rather detailed view of their own knowledge domain concerning a part of the problem and some commonsense knowledge of the whole problem. By means of a mutual exchange of knowledge, the problem can be analyzed.

We saw that handling complex interdisciplinary societal problems fruitfully requires knowledge of different phenomena. This knowledge can be found in different disciplines which have each gathered a body of knowledge about particular phenomena and their relations.

In order to collect the knowledge and data fruitfully, one has to be familiar with the domain. Domain related knowledge and data can only fruitfully be collected by persons who have studied these domains. Data can be gathered via observations based on experiment, through measurements and by reading or asking questions. This can be done in a strictly scientific way or casually by coincidence or intuition. Knowledge and data about phenomena can also be gathered from everyday experience.

Social science has developed methods for the reliable collection of data and knowledge by observation, experiment and study of the literature. Domain related knowledge and data from a single discipline must be gathered and combined with the domain related knowledge and data from other disciplines. In many of these knowledge and data collecting techniques there is a main role for the researcher who collects and analyzes the material. This, however, gives the researcher a very important role, sometimes too important. The researcher as initiator, collector and interpreter is not always the most adequate person to analyze all the aspects of a complex interdisciplinary societal problem. The researcher selects the questions based on his or her mental idea of the problem, collects the data and subsequently interprets the information based on these data, according to his or her own theoretical concepts and view of the problem, summarizing this in a report. Women's studies made it clear that the influence of the researcher, in some cases, is crucial, because one analyzes a problem on the basis of one's own theoretical concepts and prejudices[16]. Some attempts have been made to diminish the personal influence of the researcher concerning the analysis of the data, especially with complex interdisciplinary societal problems. In the case of uncertain knowledge and data, it is productive to let experts themselves interpret the data. A well known and often used method for this subject is the Delphi method developed by Dalkey (1972), which confronts experts with the opinion of the other experts. The methods that are known by the name Delphi method all use intercollegiate reflections. Delphi is a concept that applies to several methods that have more or less the same characteristics. In the Delphi method, experts are asked to answer structured questions and then comment on the answers of colleagues to these questions. There can be several rounds of comments often concluded with a final face to face discussion.

The goal of the Delphi method is to maintain the advantages of group discussion and reduce the disadvantages. The advantages of team discussion is that the mutual influence of the team members on each other leads to a team discussion reseult which is better than the sum of seperate expertise. The disadvantages are the differences in influence in the process due to difference in status, gender, power and verbal competence. The disadvantages can be reduced by structuring the often unstructured face-to-face meetings and through anonymous communication.

An advantage of gathering information in this way is that the experts do not have to travel, except for the final discussion.

Van Dijk (1993) defines research based on the Delphi method (p. 3.18) as follows:

"Delphi-research uses methods of questioning, that are based on interactive communication between members of a team (panel), who do not have to meet each other. Instead a central medium (= researchers and instruments) processes and feedsback this communication to the team ( in several rounds) as often as necessary till the goal of the research (mostly future prediction and /or policy defining (strategy) is achieved."

However the personal influence of the researcher still plays an important role in the selection of the experts and in formulating the questions.

An important point in gathering and combining knowledge and data for complex interdisciplinary societal problems is whether to let the researcher be the co-ordinator and central person in the research, or the experts themselves. As stated in chapter three, the way a problem is approached depends on a person's background, the discipline studied, personal capacities etc., all of which influence the mental idea the person has of the problem and on the basis of the mental idea the problem will be approached. Observations are made on the basis of this mental idea, the literature is selected, the questions are asked and new data interpreted and combined.

What approach is the most fruitful for handling complex interdisciplinary problems depends on how much knowledge and data already exist about the problem and in what phase the problem is. One's selection of an approach will depend on how much a problem is structured and what kind of information is required. When many things are known and only a small amount of complimentary knowledge and data are needed, it may suffice to ask experts individually. When there are nasty issues that can easily be formulated into questions, and on which intercollegiate reflection would be fruitful, the Delphi method may be the correct methodology. However, many complex interdisciplinary problems remain undefined and are often very vague. For problems in this situation no structural questions can be asked. In such cases it will be more fruitful to give experts the opportunity to discuss the problem together in a mutual co-operative process of problem definition. The direct contact with each other gives experts the opportunity to ask for further information, to brainstorm together, to ask questions, to explain the special domain knowledge to each other, to express doubts, to speak privately, to utter criticism, to make decisions together, to formulate the models and the problem together and to discuss interventions and scenarios. This can be combined with various methods of information gathering. One of the main reasons for preferring co-operative problem handling is that the experts should handle the problem together. The problem is often too complicated to be handled single, there are always new questions and new data that need to be correctly interpreted. This can only be done by the experts themselves. They are the ones that can combine the knowledge needed to define and analyze the problem. Together they must try to define the problem by filling the remaining knowledge gaps and forming a picture of the problem, as far as this is possible, from the combination of the pieces of domain knowledge. Interaction in co-operative problem handling gives a higher dimension to the process of information exchange. Interaction not only fills some of the knowledge gaps of the fields other than the speciality of the expert, but can also stimulate one's own insights in one's own field.

What is different about this way of information exchange compared with other forms of information exchange described earlier? In other problem analyzing methods the researcher is the sole co-ordinator of the knowledge and everything is arranged according to his or her interpretation. In many cases the experts are unable to interact with each other, nor are they informed of each others knowledge on the subject. In the Delphi method, although experts are informed about the knowledge of others, this is only a very limited and restricted kind of interaction. Besides, the Delphi method only supports inter-collegiate reflection on a few moments in the problem handling process. A method for analyzing and defining complex interdisciplinary societal problems should primarily support the interaction between experts. Other aspects required, how and if negative group aspects can be avoided and which tools can support this process, will be described in the next section.

A fruitful way to define a complex interdisciplinary societal problem is letting experts, each with their own knowledge concerning the problem, handle the problem in co-operation. However, co-operative problem handling has positive and negative aspects. We will start with the positive aspects.

We prefer the experts to be in direct confrontation with each other. In the co-operative problem handling process there is a continuous interaction of one's own mental ideas about the problem and those of others. This mutual interaction makes it possible to change the mental ideas of all the participants of the problem handling process. In this confrontation and in the opportunity to share the knowledge with each other, the personal domain knowledge about phenomena of the problem can be combined to produce a collective quantity of knowledge from different domains[17].

Even at the first exchange of information the mental ideas about the problem of the participants[18] will change, and this process will continue through all phases of the problem handling process. There is not only an intra-personal change influenced by reading, discussing, thinking and data gathering as is described in chapter three, but also an interpersonal exchange, which will ensure that the outcome of the process differs from the process of interviewing people individually, even when it is combined with the Delphi method. Defining complex interdisciplinary societal problems, and making a model and for suggesting interventions, requires creativity, discussions, brainstorming, and information exchange by people with knowledge of the main aspects of the problem.

Co-operative problem handling is a sine qua non for the analysis and definition of complex interdisciplinary societal problems. The process of co-operative problem handling with a group of participants[19] will not always be as smooth and co-operative as the initiator wants it to be.

As described earlier in chapter three, there may be power differences between the participants due to achieved status and/or ascribed status. These achieved and ascribed status differences can influence the process negatively. Achieved status includes being a famous person or belonging to a more 'important' or 'more highly valued' profession or at a higher level in the same profession than others[20]. Ascribed status includes gender, skin color or character. Because each participant is invited on the basis of her or his expertise and all expertise is equally necessary for the success of the problem handling, it is very important that each participant has equal opportunity to participate actively, to give information and to influence the process. Achieved and ascribed factors can influence these opportunities negatively.

Other things that influence the information exchange negatively are, attractions and antagonism between persons and mutual conflicts. Jealousy and competition can cause people to withhold information.

Another factor that can influence co-operation negatively is hidden agendas. One cannot assume that everybody's goal is to achieve the best result, quite apart from the fact that it is not clear what the best result is, for whom it is the best result, and who decides what the best result is. Different political views, divergent goals, and different personal interests can interfere negatively.

Another difficult point is that participants in a multi-disciplinary team come from different fields and different backgrounds.

All these aspects can cause serious communication problems and prevent an optimal group co-operation, in turn preventing an optimal result.

A method for guiding the co-operative group process should stimulate the positive aspects and avoid the negative aspects. In reflecting this we are aware that it is not always possible to avoid all negative aspects. However, it should be attempted.

To a certain extent, differences in power can be avoided in written communication when people are anonymous. Written information is less open to influence by internal and external factors. However even in evaluating the written information and interpreting the knowledge, power differences cannot be totally abolished. To avoid connecting written information to persons, one could ask the participants to submit their information anonymously. This is one of the main planks of the Delphi method, where the information is collected and aggregated[21]. This only works with large groups or when participants have not met before, since otherwise a person can easily be traced back. On the other hand, anonymous information can lead to difficulties in interpretation of the information. In our view, all information is colored by knowing one's political point of view, historical background and interest. Since we assume that there is no such thing as value free knowledge (Weber, 1922), or power free knowledge (Habermas, 1972), the information given by a person can be better valued if these values are made explicit. In verbal communication it is more difficult to avoid this problem. When the group of participants actually meet, gender and skin color is soon recognized.

How to avoid the negative aspects of influences and meanwhile inform each other about personal backgrounds? Where a bulk of written information is suitable, for instance, at the start of the problem handling process where participants describe their professional view of the problem, we would prefer written information over direct confrontation. Here we would be satisfied with information about the professional background only.

How to handle problems of the retention of information and hidden agendas, is another matter. These are sometimes very hard to detect and/or to avoid. It is not always clear to see when people are holding back information or when they have hidden agendas. Where it is obvious, the facilitator[22] should try to discuss this with the person concerned and remind him or her of the mutual goal of the group process, which he or she subscribes to. Where the person concerned is not able or willing to change, as a last resort the person can be replaced by another person with similar expertise. Conflicts or jealousy can be discussed either in private with the facilitator, with the one or two persons involved, or with the whole group. But one should be very reticent. One should avoid emotional group processes or becoming too personal. One should not forget that it is a professional group with professional goals. The goal is not to explore one's feelings, but to explore what the problem looks like. In case of conflicts the participants can be reminded of the goal of their assignment and can be asked if they intend to act in a responsible and motivated way towards their tasks. In this, one should follow a professional approach rather as Weber (1922) described for the civil servant: in working together without letting one's personal feelings interfere negatively.

It is difficult to avoid the negative aspects of group processes. To what extent this can be avoided should be analyzed further in empirical research. Some things can be implemented in the method, some things can be taken care of by a well-trained guide of the process, while other things just cannot be avoided. Being aware that this can happen is already one step on the way to avoiding it.

A well trained and experienced group facilitator should not be afraid to intervene when factors like dominance, jealousy, hidden agendas and obvious hostility arise.

'Group think'[23] is another intervening variable often mentioned, and which prevents an optimal result. The term 'group think' comes from the psychologist Janis, from his psychological theory of failing decision making. He defines group think as (Janis, 1972, p. 9):

"A mode of thinking that people engage in when they are deeply involved in a cohesive in-group, when the members' striving for unanimity override their motivation to realistically appraise alternative courses of action".

The differences in power described in the previous section can lead to group think. Rosenthal (1984, p. 64-65) calls 'group think' a process of which decision makers come to a collective definition of a situation or to a unanimous decision of recommendation and avoid a well-considered confrontation with the problem. This can be caused by overestimation of oneself, collective narrowing or pressure to conform. These symptoms of inadequate decision making have only a slight chance of leading to a good result. Decisions will be based on short-sighted conformism. 'Group think' leads to decisions that do not take into account the chances of serious risks and dangers. 'Group think' can result in underestimating the risks of the strategy to follow. 't Hart (1991b) sees two variants of group think: the first as the collective avoidance of problems by groups of decision makers under stress, which is Janis' original meaning of the term, and 'group think' as an over-optimistic thinking of groups involved in decision making or working on projects that they consider as big challenges ('t Hart, 1990).

- internal group processes, in which the individual loses control over his or her own thinking and handling;

- social control as a result of which the individual no longer dares to state his or her different opinion;

- anticipation of the reaction of superiors, by which people tend to want a collective cover against criticism.

'Group think' can characteristically occur in cohesive groups, belonging to an organization with structural failure, that are handling problems in a provocative contextual situation. 'Group think' often occurs in crisis (Rosenthal, 1984). Examples of this are the Chamberlains inner circle between 1937-1938, in the US government concerning Pearl Harbor in 1941 (Allison, 1971, p. 203-205; Rosenthal. 1984, p. 60) and concerning Cuba with the Bay of Pigs invasion in 1961(Janis, 1972; Rosenthal, 1984, p. 40, 60, 70-71; Allison, 1971, p. 205; Halperin, 1972, 147-148).

Rosenthal gives some heuristics for avoiding the group think processes and to neutralize their consequences. These are[24]:

7. taking time to make up an inventory and judge the risks of alternative scenarios;

8. special meetings after a critical decision in with each participant takes the role of the devil's advocate[26].

What kind of implications can this have for our approach to co-operative problem handling? What should be done and what should be avoided? If we follow Rosenthal's suggestion for avoiding group think this means that:

1. the facilitator and, maybe, also the participants should reward the utterance of objections and doubts;

3. at crucial moments in the process of defining one should form parallel ad hoc groups and compare the results of the groups with each other;

5. periodically, outsiders should be invited to critically observe the group process (DeTombe, 1993b);

6. the facilitator should invite different persons of the group to play the role

of the devil's advocate. When everybody takes this role alternately, the negative aspects of the role of the devil's advocate can be more or less neutralized[27];

7. there should be time to make up an inventory and see what has been forgotten and time to make alternative scenarios;

8. there should be special meetings after critical decisions in which each participant takes on the role of the devil's advocate.

These heuristics are time consuming and do not make the decision process easier. But given the problems one must take time to do this. These problems and the consequences they have for many people are too important.

Another thing that can interfere with an optimal handling are collective blind spots[28]. By collective blind spots, we mean phenomena that all participants of the group forget to include in the description of the problem. These can be very common things which one simply overlooks, minor aspects, but it can also happen that one overlooks whole sectors, or groups of people, that actually influence the problem or are influenced by the problem. Because of the complexity involved, this can often happen in analyzing complex interdisciplinary societal problems, and is very difficult to avoid. Collective blind spots are more or less inherent to complexity.

One of the things that, can serve to help avoid collective blind spots, is breaking through the collectivism of the group handling process, either by a continual altenation of co-operative problem defining and individual problem defining, by frequently asking the opinion of colleagues outside the group process and or by forming parallel groups and inviting, periodically, outsiders to observe and provide criticism[29]. Another thing that helps prevent collective blind spots, to some extent, is constantly making participants aware of the possibility of these collective blind spots.

In an interdisciplinary team each person has his or her own profession and background. Experts often use a professional language, in to explain things and talk about their own field. Professional language is specially developed to make it possible to express oneself clearly within one's profession. In this vocabulary many concepts and phenomena are very well defined. However, each profession has its own vocabulary, which does not belong to the set of common knowledge everybody has. In an interdisciplinary team one should carefully indicate where the ideas come from and what the concepts mean in order to avoid misconceptions and excessive verbalism. A word in one profession can have a totally different connotation in another profession. Explicating all the concepts clearly will prevent the participants from using verbalism and will make it easier to realize where some parts of the description of the problem come from. In order to communicate well it is advisable to avoid professional terms. Should this be impossible, one should describe a limited number of professional terms and explain the meaning as well as possible. We are aware that it will not be easy to break the habit of using professional concepts and words.

In order to stimulate and guide the information exchange, the group must be guided, possibly by a chairman. A chairman, when performing well, can stimulate and co-ordinate the information exchange in a group. The role of a chairman can vary from a person who only guides the group process and does not interfere with the goals, i.e. a neutral chairman, to a chairman who directs the groups towards his or her own ideas. Handling complex interdisciplinary societal problems, however, requires more than just some normal meetings. This process must be guided more carefully and for this a special method is needed. The process should be guided by a specially trained person, who is acquainted with the method and tools. To distinguish this role from that of a chairman, the person is called a facilitator.

The task of the facilitator is to guide the co-operative problem handling process. This should be an experienced person, who has a reasonable amount of psychological insight into problem handling and into group processes. The task of the facilitator is to guide the group towards its goals, by stimulating the information exchange and seeing to it that every member of the group can participate fruitfully in the discussion. The facilitator must be aware of and should avoid those negative aspects of co-operative problem handling mentioned above. A method that supports complex interdisciplinary societal problems can only prescribe to a certain extent the problem handling steps and moments. Sometimes it can be more productive to deviate a little from the method. In order to know where and when this is possible, the facilitator has to be a highly qualified and experienced person, who is alert and who will feel free to improvise when necessary.

There are moments in the problem handling process when the computer can be of assistance. Although a computer assistant can help the facilitator in this, the facilitator him or herself should also be skilled at handling the software in order to give directions, to know what is possible or to prepare or try out things on his or her own. The facilitator should also be well acquainted with a range of social science methods of data retrieval and information collection and have a good view of where these supplementary methods are needed and can be implemented during the problem handling process.

The role of the facilitator should be neutral and impartial towards the outcome of the process, also to prevent 'group think'. A facilitator should not interfere in the content, he or she only controls and guides the process of the information exchange and group process. The work of the facilitator consumes a reasonable amount of time during the problem handling process[30].

An interesting question is "How much should the facilitator know about the problem itself?". As was explained earlier, it is impossible for one person to know all the aspects of a complex interdisciplinary societal problem, and this includes the facilitator. However the facilitator should at least have some common sense knowledge of the subject at the level of what is published in a quality paper about this subject. Furthermore the facilitator should prepare him or herself for the specific topic by some exploratory talk and in depth interviews with some content experts. The facilitator should at least, to some extent, understand the problem in order to be able to guide the process fruitfully.

As we have argued before, handling complex interdisciplinary societal problems is not a one person job. Selecting a team already includes or excludes certain 'solutions'. The selection is therefore a very important step in the process of problem handling.

Who should do the selection? In the case of a problem of knowledge or policy a committee or the (local) government can invite people. This is often done by giving an assignment to a special assignment bureau[31]. The invitation could also come from a special interest group, or could be done by the university, a large organization, an action group or a pressure group.

Who should be selected? What are the criteria for selection? Should one select persons with general, overall knowledge about the problem, or should one select persons with detailed specialist knowledge about the problem: experts. What expertise is required in the team depends on what is already known about the problem, the urgency, the importance and the political impact of the problem.

Selecting a team is best done on the basis of the mental idea of the problem. Sometimes a mixed group of content experts and political experts is a fruitful combination for handling policy problems. Another possibility is to let content experts first analyze and define the problem and suggest some interventions, and then let a group of politicians discuss the political consequences and possibilities.

When it is primarily a problem of knowledge, which is the case in many new complex interdisciplinary societal problems, experts on the domains involved should be invited to analyze the problem. Where many phenomena of the problem are already known, but political support is missing, representatives of political power groups can be selected. In the case of important changes within an organization, it may be wise to let employees of the organization analyze the problem themselves, since the outcome of the process will have a greater chance of being accepted, if it is considered their own idea (Vennix, 1990). In the case of a conflict, when there is more than one party involved, representatives of the different parties should be asked (Glasbergen, 1992). Glasbergen defines co-operative handling of the problem by different parties which are involved in the problem, network-policy. During the process of problem handling it may be necessary to make some changes in the composition of the group. Choosing people and selecting the domains can be a circular process. While clarifying which domains are involved and what is important in the problem, some experts may leave the group, while others may join the group.

In composing the group, one has to be aware that there are limits to the group size. A great deal has been written about group sizes. On the one hand the group size should not be too large and on the other hand it should not be too small (Hart, de Jong & Korsten, 1991, pp. 244-246). It is not clear in advance how many participants are needed. It depends on the complexity of the problem and whether people are available. The number of people depends on the amount of different domains that are influenced by the problem. Although we cannot indicate the optimal group size beforehand, we can say something about the minimum or maximum group size. In order to make fruitful mutual communication possible, the group should not be larger than twelve to fifteen persons. On the other hand, there should be enough people to discuss and exchange opinions and there should be enough people to share responsibility. Therefore, the minimum group size would be something like five to seven.

Where many domains are involved and therefore the group is going to be too large, one could:

- form a team with knowledge experts with the expertise in the main domains of the problem and in later group session, incidentally ask people with more specialized knowledge on details of the problem.

- for the first problem defining sessions select people with general knowledge of the problem, and have this be followed by a team of experts.

- begin with some of the experts and later on change some of the experts in the team. However, this will have consequences for the team spirit.

There can be an important differences in the time required to handle different problems. Some of these time differences have already been mentioned in chapter three, concerning riots and hostages (Rosenthal, 1984)[32]. Complex interdisciplinary societal problems like those need immediate handling, with very little time to discuss the domains, to handle the problem carefully, or to select a team. Moreover the interventions often have major societal consequences. Making the wrong decisions could have a great negative impact. When intervening under stress, the danger of 'group think is greater than in other situations. Under stress people tend to say there is no time for discussion and then take the decision themselves or give other people the full responsibility instead of dividing the responsibilities among all the members of the team (Rosenthal, 1984).

When nobody has the answer, and when there is panic and pressure of time, it is conceivable that the only suggestion given will be followed. However, even when working under the pressure of time, one should define the problem carefully and discuss the consequences of interventions carefully. Because of the important consequences the interventions can have, the idea of Rosenthal (1984) of forming a parallel group to analyze the same problem and discuss the interventions, before acting, is a good way of preventing group think and other kinds of errors.

Some urgent problems change rapidly, sometimes within hours, making it very difficult to guide them. Because of the little time available in these cases, one should, as far as this is possible, be prepared for these situations in advance. One should have several teams of specialists from different disciplines ready, spread over the regions, who are familiar with and regularly trained for the analysis of all kinds of complex interdisciplinary societal problems under time pressure.

Handling these kinds of problems must take place in at least two phases: first, interventions in order to save lives that may be in immediate danger; later, when the immediate danger is at least temporarily over, a reconsideration of the whole problem as such in a more relaxed atmosphere, with the analysis and action towards the problem further. Then there is time to reconsider and analyze the problem more thoroughly, with the focus especially on primary prevention.

In the first stage of handling a problem, the problem should first be defined. The phases of problem definition are described in chapter three. Co-operative problem handling often starts at the moment when the first phase of the first sub-cycle of the problem handling process is completed[33], when there is a vague mental idea formed about the problem. This can be developed individually, by a group of persons or by an action or a pressure group[34].

This vague mental idea about the problem is often different for each participant because it is influenced by the different profession and background of each participant. These individual mental ideas can be put into words, and these descriptions combined into an overall description of the problem. This description can be the beginning of the mutual co-operative problem handling process. In words, one can (clearly) express what causes the problem, what are the effects of the problem and which phenomena influence each other, why phenomena influence each other, based on which theoretical ideas and empirical data. Step by step one can extend the description of the problem in an iterating process of combining, linking the mental ideas of the participants to data and by collecting data and changing the mental idea. The problem is described on the basis of empirical data and on theoretical ideas about how and why the phenomena that are involved are connected. The theoretical ideas can be based on theories, hypotheses, assumptions experience or intuition. In proceeding from a vague mental idea of the problem to constructing the conceptual model it is important to know in the context of which theoretical ideas[35] the data have been gathered in order to be able to indicate the status of these data. How should the data be considered, based on what theoretical ideas are the phenomena influencing one another. These theoretical ideas should be carefully described for two reasons.

First, when in the process of data gathering, some data are in contradiction with other data, it is important to know on what the ideas of connecting the phenomena are based and what could be the reason for discrepancies. This can be a reason for further (re)search.

Second, because there are participants from different backgrounds each participant has his or her own professional language, own theoretical ideas, own concepts and concern with particular phenomena. These professional concepts must be used very carefully in the description of the problem. In order to prevent misconceptions each professional concept, and each phenomenon, should be carefully defined. How phenomena are related to each other can be expressed graphically in a semantic and in a causal model[36]. White spots and blind spots can be indicated by drawing the knowledge islands.

Describing the problem in words, stating on what theoretical ideas the description is based and making the semantic model and the causal model of the problem is an iterative process, as a result of which the conceptual model can be defined[37].

On the basis of the verbal description of the problem one can construct a semantic model in which relations between the phenomena are connected. First this can be done in words, then, on the basis of this description, a graphic representation of the connection can be made and when a problem can be analyzed in terms of cause and effect, a graphic representation based on cause and effect can be made. The description of the problem can be modified according to the graphic representation in the semantic model.

As stated earlier[38], expressing the problems in different languages can help to give a clearer view of the problem. This is the reason why these steps from verbal description of a theory to defining concepts to knowledge islands to semantic networks to causal networks is not a linear process but an iterative process. What is expressed in one model described in a certain language, can be added to another model, and described in another language.

As explained earlier, complex interdisciplinary societal problems should be defined primarily in co-operation. This should be alternated regularly with individual preparations. Individual preparations are needed to be able to prepare a paper or a lecture about the domain part of the problem, to be able to think and contemplate without being disturbed by others and to be able to refer to the literature and other documents.

In addition to mutual sessions and individual preparations there have to be opportunities where other domain specialists can be consulted. Because there is considerable uncertainty and new issues arise concerning these kinds of problems, it is sometimes necessary for participants to consult someone from their own domain. They can discuss specific details exhaustively. This domain group can also serve as a group of critical observers, and in this way the group can also be virtually extended.

Another point of interest is that co-operatively constructing a model will make it easier to have the model accepted as an expression of the problem by those who have developed the model. This way of making models can serve as a means to smooth the acceptation of changes by a group of persons in an organization. The persons who have developed the model have the idea that the image of the problem expressed in the model is their own 'solution' and therefore will accept the changes more readily(Vennix, 1990; Wierda, 1991). This can be useful in conflict situations. On the negative side, in working towards a consensus in making the model one easily provokes 'group think'. When a model of a problem is not accepted by someone there must be reasons for this. It is interesting to find out what these reasons are. These may be personal and political reasons, but it may also be the case that the person has quite a different view on the problem. In the case of a knowledge problem, and those are the problems we primarily focus on in this study, one should not strive for consensus at all costs. In the view of blind spots and white spots it is interesting to find out what the reason for not accepting the model is. However, in order to proceed to the next sub-cycle, the sub-cycle of changes, there must be some kind of acceptation of the model, or otherwise the group will not be able to analyze the problem further.

7.4 Computer tools to support analysis and definition of complex interdisciplinary societal problems

What tools can can be used to assist the process of construction of a definition of the problem?

In chapters four, five and six we discussed some possibilities of computer programs for replacing or assisting the human being in handling complex interdisciplinary societal problems. We concluded that Artificial Intelligence was of limited usefulness[39].

However, conventional programs can be useful in applying general application programs, for instance, databases and text writers. Some ideas about group decision support programs in particular seem to be challenging. The computer can also support the different models for expressing the conceptual model of the problem. For the description of the problem in natural language, the computer can be a good medium. During discussion, the description can be easily adapted, and so can the concepts and the description of the phenomena and the theoretical ideas. A computer tool that supports this process must be able to adjust the text easily during the process of defining and must be able to make, in a simple way, the connection between the words in the text and the concepts and description of the phenomena and the theoretical ideas[40]. When one alters one thing one should check whether the description of this issue also has to be changed in the other descriptions of the problem.

When changing the description of the problem, one must not forget to change the relations in the semantic network, and in the theoretical description. To prevent this, the phenomena are marked, for instance, with a character that corresponds to the other descriptions. Different computer screens can be connected in this way by hypertext. There are several software packages that can support hypertext, graphical drawings, normal text[41]. By means of hypertext, one can connect the description of the problem with the definition of the concepts and the concepts with the theoretical ideas. This can be connected to the semantic model and to the causal model[42].

A simulation model based on differential equations is a correct tool for expressing changes over time. By making models of complex interdisciplinary societal problems, Forrester (1987) and Meadows (1980) recognized that handling differential equations directly can be very difficult for non-mathematicians. In order to be able to include non-mathematicians in the process of modeling they developed a simulation software package in which mathematical relations between phenomena can be directly, graphically expressed, and then automatically translated by the software into differential equations. This combination is implemented in system dynamic simulation software, where a graphic representation is combined with a mathematical representation to express changes over time. System dynamic modeling software is based on the ideas of system theory[43] and further developed by Forrester (1987/90).

In the system dynamic modeling one can express the mathematical relation between phenomena graphically. Forrester made it possible to translate the graphical relation into the mathematical formulae that describe the relation between phenomena. In the software, the phenomena are represented by circles or squares and the influences are indicated by arrows, while the relations between the variables are indicated by equations. The model uses differential equations to indicate the change of the variables over time. System dynamic modelling is graphically expressed in a flow diagram, in which the influences, the cause and the effect of the phenomena on each other are expressed by flow charts. These influences on the phenomena can be expressed in the so-called help-variables.

The characteristics of the elements that system dynamics uses for expressing a problem (Flood, 1992, p. 63) are: order, direction of feedback, non-linearity, loop multiplicity.

Order is expressed in the number of 'levels' that are used to represent the structure. These are the levels that represent an amount of something, such as an amount of animals, or amount of manure (in system dynamics this can go to nearly one hundred elements).

The direction of feedback can be from one element to another (direct or indirect), and can be either negative or positive. Negative means an inhibitory or controlling influence, positive means an augmenting influence.

Non-linear systems influenced by positive feedback produce exponential growth or decline from a set point. Loop multiplicity means that, normally, more than one loop structure is needed to express the situation.

The relationships between the elements are flows termed "rates", that connect the main elements, termed "levels". In Flood's (1992) definition (p. 68):

"A level is a quantity that dynamically changes over time and a rate is a flow which contributes to the change per unit of time."

Levels and rates are necessary to describe a situation. Auxiliaries are used to decompose rate equations into manageable parts, to convert flows from one type to another, and to provide information to change/control rates, which is particularly useful.

In system dynamic modeling there are input and output variables in the state variable. These input and output variables can influence other state variables or help variables.

In building a model, the first issue in system dynamics is deciding which 'order of the system' should be developed. This is based on the semantic and causal model where the relevant phenomena are already expressed via a causal loop diagram. A signed digraph, or in our case a causal loop diagram in a causal model, is an aid for making a system dynamic is.

A signed digraph is an expression of how elements influence and interact with other elements converted to a system dynamic flow. This can be converted to a system dynamic flow diagram.

Although system dynamics was basically developed to express changes and to explore changes, the easy way of building a graphic representation makes this instrument a useful tool for focusing the mental ideas on to the same issue. Changes in time are irreversible, this means that one can indicate changes in causes and effects. This does not mean that the causes and effects have only one direction. Often in these complex interdisciplinary societal problems, variables have a mutual influence on each other. But most of these mutual influences can be analyzed in (small) discrete time steps, like two people who interact with each other.

The causal model can be the start of the system dynamic model. The 'knots' in the network can be expressed in levels and variables of the model. In this more exact expression of (a part) of the problem one can encounter issues that force one to go back to the description again.

In system dynamic modeling the model is constructed via a verbal description of the problem and via a semantic and a causal model. There are special computer software packages that support system dynamic modeling[45].

Through constant iteration of expressing the problem in words, defining the concepts, making the semantic model, describing the theory and system dynamic modeling one can form the model. This continuous shift from words to graphical model and differential equations prevent the communication from becoming too vague, not clear what one is talking about and the use of a magic language.

A simple way for starting a system dynamic model is to start with a linear model. and from there on to indicate the mutual relations. This can result in non-linear relations that can be described by chaos theoretical concepts[46].

To define the problem, the simulation model, without the quantification of the variables will be adequate for a conceptualization and for the discussion. For the construction of the conceptual model of the problem, the semantic, causal and system dynamic simulation models will be primarily used as vehicles for the discussion. Defining the conceptual model of the problem is not the end of handling the problem. A tool that can also express changes is needed to explore future developments.

In this case the conceptual model can provide the start for the empirical model and for developing scenarios. In the empirical model the variables can be quantified.

Although developed for organizations, this idea for modeling is also used for other issues. System dynamics uses, for simulation modeling, time based mathematical equations for predicting of future development. Forrester has developed a theory of information feedback and content as a means of evaluating business and other organizational and social contexts. These ideas also involve an analytical modeling methodology. One idea is that any situation can be considered as complex, but mainly in terms of elements and "flows". Flows are the relationships between the elements. All influential elements must be included within a boundary. Connecting elements may form loops and hence feedback analysis is considered as very important (Flood, 1992, p. 61).

"It is a simple use of standard traditional systems ideas, called system theory, that have been developed into the elegant mathematics of linear control theory and the much more troublesome non-linear control theory in other domains, such as control engineering. Its accessibility, however, makes SD (System Dynamics[47]) an approach that can be understood and/or drawn upon those in organisations involved in decision making."

As indicated earlier, System Dynamics is based on standard systems theory, which developed into the use of system dynamics for analysis. It is assumed that social reality is best considered simple and unitary (Flood, 1992).

"Although there may be many elements and relationships, the models structures developed are normally deterministic and do not evolve over time (the parameters are time invariant), and the model is largely closed to the environment (all influential factors are contained within the main inputs being in the form of perturbations to the model). Analysis is, therefore, along a closed sequence of causes and effects, a closed path of action and information."

Flood states (1992, p. 62) that a system dynamics view emphasises on structure and the processes within that structure, assuming that this is how dynamic behavior in the 'real' world can best be characterized.

We will not go that far. We consider that a system dynamic model is one of the many ways a problem can be expressed. However, in this study it is used more, as a discussion vehicle, as an open system in time. System dynamics considers behavior principally caused by structure. This assumes that the problem can be analyzed from an external, objective viewpoint, that the 'real' world can be re-created in the model.

Although we use system dynamic methodology, only to a certain extent do we subscribe to the ideas underlying the theoretical ideas of system dynamics. We do not subscribe to the idea that a problem can be expressed from an objective viewpoint. At most, we think that it is possible to create a view from a mutual viewpoint of a team of persons, an inter personal viewpoint. Representing a problem in terms of phenomena with output and input and the relation between the phenomena as feedback loops in a dynamic system is a step towards coding this representation into a computer program, which can simulate the interactions and as a consequence create future business scenarios.

Although we do not totally subscribe to the ideas of system dynamics, we do find the way that a problem can be modeled by system dynamic methodology and prepared as a simulation model a fruitful way in the first sub-cycle of problem handling, as a vehicle for discussion, and in the second sub-cycle of problem handling as a tool for expressing the empirical model, as a vehicle for discussing interventions and as a tool for constructing scenarios.

The software[48], developed on this idea of system dynamic modeling, uses the concept/ idea of object oriented modeling. There can be several advantages in using a system dynamic software tool:

With this tool the facilitator (directed by the group) can prepare a graphic representation of the conceptual model of the problem. The easy on-going building of a model with this tool makes it possible to form a model during the discussion. The tool can work on a small part of the problem while neglecting other parts. It does not necessarily need very detailed or complete data. One can also work with estimates of data left out or add new variables to the model.

The working memory of the human being is very limited being as it is, in the active data it can hold (Crombag, 1984). This tool helps to store the data and makes it possible to maintain an overview of the whole model. The graphic presentation and the ability to direct to some points in the model helps the group to focus on the same issue. The system dynamic model can be a vehicle for focusing the discussion in the interdisciplinary team. The system dynamic software can serve as a mutual language for the participants during the conceptualization of the problem.

The problem handling process can be supported with all kinds of regular computer support. We mention briefly a computer tool for brainstorming, for arranging data and for computing.

The idea of brainstorming developed by Osborn (1953), originated in the field of advertisement. The idea is to amplify the creativity of the individuals. Creativity and new ideas are evoked by giving a group of people the opportunity to utter new and creative ideas, and in so doing stimulate other to come up with new ideas[49]. Although this situation is intended to reduce self-criticism and criticism of others, these aims are not always meet. Because this is mostly a verbal event, not everybody feels free to utter all kinds of crazy ideas. The input of each member in the group process can be very different because of personal dominance. At some moments in co-operative problem handling, power differences can have a great influence, as in at moments of voting and brainstorming. If we want all the participants to have an equal influence in this process, references to persons should be avoided. In these cases anonymity must be considered, alternated with open discussion. One can use a computer program specialized in anonymous data collection for voting and brainstorming. However some statements/sentences can still be traced back to a particular person. For example, one recognizes the person, by the subject or because of the formulation of the sentences or the use of specific concepts or words.

To handle these new problems that have often never been handled before, brainstorming is a useful technique for triggering all kinds of ideas. When carried out well, brainstorming can be a fruitful technique for generating all kinds of ideas, creative ideas, correct and incorrect ideas, simple ideas, strange ideas, fruitful ideas. Brainstorming can be used at several moments in the problem handling process. Therefore it is important that, right from the start of the process, this technique is used optimally. For analyzing complex interdisciplinary societal problems much inventiveness, creativity is often necessary. With brainstorming, one does not need to take the specific form of the problem into account, including constraints of reality. There is space/freedom in the problem, and this space/freedom can lead to new solutions.

For brainstorming one can use a computer. This has several advantages. Each participant can type his or her own ideas into the computer without being disturbed or embarrassed by the comments of the other participants. Using the computer for comment can also be a way of avoiding 'group think'. The computer can collect, sort, and rearrange the outcome and help as a collector of ideas, so that no ideas will be omitted. The computer can print out the sorted data, which can then be the basis for further group discussion.

Brainstorming steps: each person has one computer. The facilitator has a file-server which can collect and distribute the computer data. The facilitator stimulates the participants to make a list on the computer of all the possible aspects, countries etc. that are, in one way or another, connected to the problem. These things need not be ordered. The participants can, at random, just type in their ideas. This material is sent to the central computer. This the material can then be categorized by the facilitator with the help of the participants. The participants can be asked to indicate how, and why these things are connected with the problem. While doing so, they can add more issues to the problem connecting list. The ideas of others and the categorization of the material can help to trigger new ideas to the participants. Using special group ware such as, for example, Group System V, the whole process of brainstorming can be carried out anonymously[50]. Participants can be asked to comment, also anonymously, on the material of others, give additions and utter criticism. After categorizing, the ideas can be discussed in a plenary session. Individual participants can be asked to explain their ideas or their comments, either openly or anonymously. At different moments in the process of problem handling, voting, selecting and ranking of material, uttering criticism or playing the devil's advocate can be introduced. Here, anonymity is also welcome to prevent 'group think'. This part can also be supported by group ware, in this case Group System V[51].

As indicated in chapter three, much data needed for analyzing complex interdisciplinary societal problems is often missing, or will take much time to collect. Therefore existing data bases can be consulted. For instance, data from the central bureau of statistics (the census bureau), a database for research literature can be used, databases for journals, and some of the special professional databases which gather professional data. In the process of analysis, participants must make a list of all the phenomena connected with the problem and that have to be categorized. This list can be constructed in a special database. This can be done in any kind of database software program as long as the data can be easily converted from one software package to another.

There are often moments when calculations have to be made, and then a spreadsheet is a good tool to use. This can be any kind of spreadsheet as long as it is simple to handle.

Facilities such as, for example, print and fax facilities, telephone lines and electronic mail also belong to the normal outfit of technological support for handling complex interdisciplinary societal problems. The telephone, fax or electronic mail can either be used by the participants or by the facilitator. For some problems or some moments in the problem handling process it may be that participants can only be virtually present. Then two groups of participants, each in a different location, can be connected through tele-conferring or video conferring.

7.5 A list of conditions for a method for analyzing and defining a complex interdisciplinary societal problem

To conclude the earlier discussion of analyzing and defining complex interdisciplinary societal problems, we are now able to state some of the conditions that must be fulfilled by a method for supporting analyzing and defining complex interdisciplinary societal problems. We are aware that we cannot prescribe every detail and therefore we limit ourselves to the main aspects already discussed in this study. This means that the list will be neither complete nor exhaustive[52].

A method that supports the definintion of a complex interdisciplinary societal problem should assist the positive aspects of co-operative problem handling, avoid the negative aspects and should guide the change from a mental idea to a conceptual model. We have already indicated that some of the negative aspects of group processes, things such as differences in power and collective blind spots, can only be avoided to a certain extent.

We are aware that it will probably be impossible to find an overall ideal method for analyzing all kinds of complex interdisciplinary societal problems, even if, as we do, we limit the description to knowledge problems, which can be analyzed in causal relations. But we are convinced that approaching a complex interdisciplinary societal problem this way gives a reasonable chance of finding a fruitful definition of the problem[53].

In the description of the method, we limit ourselves to those complex interdisciplinary societal problems which it is agreed that professionals such as, for instance, members of the government or policy makers should handle, and to those complex interdisciplinary societal problems which are primarily knowledge problems[54].

In describing the conditions to be satisfied by a method for analyzing and defining complex interdisciplinary societal problems, we will not start with the beginning of the problem handling process, the moment that one becomes aware of a problem. The process of problem finding needs a special approach, which goes beyond the limits of this study. Neither will the method support the process right through from awareness to taking the responsibility to act towards the problem[55]. The method assists in the construction of a conceptual model. Although this conceptual model can be the start of the empirical model, we will not discuss the method beyond the first sub-cycle.

The method starts at the moment when there is awareness of a complex interdisciplinary societal problem and a commitment to handle the problem. The method begins at the second phase of the first sub-cycle of problem handling (phase 1.3) and supports the problem handling up to the last phase of the first sub-cycle (phase 1.4)[56].

- awareness that each person of the team has only a partial view of the problem and a more detailed view according to his or her expertise

By naming those aspects which the method should support we summarize the issues discussed before. In this we limit ourselves to a method for supporting (part of) the first sub-cycle of problem handling. Because many researchers focus their research on the way people handle already defined problems, little attention is given to that part of the problem handling process where a problem is defined. Defining a problem is very important. There is a close connection between interventions in the problem and the definition of the problem. We have seen that complex interdisciplinary societal problems are often undefined. Before interventions can be fruitfully suggested the problem must be defined. In the method we suggest, the emphasis is on the first sub-cycle of the problem handling process: the cycle of defining the problem and constructing the conceptual model.

Handling complex interdisciplinary societal problems requires a co-operative approach. There are too many disciplines and phenomena involved to be known and overseen by one person alone. Another reason is that many of these kinds of problem are too important to be handled by any single person. A team of people from different disciplines is needed who, together, with mutual effort, handle the problem.

Many of the complex interdisciplinary societal problems are undefined. Before suggesting interventions the problem should be defined. The conceptual model of the problem is constructed in an iterative process of relating the mental idea with data from reading and observation. This process can be supported by written and verbal information exchange, partially supported by the computer. The method should support describing the phenomena, the concepts, the assumptions, hypotheses and theories, constructing a semantic, causal and simulation model of the problem.

For defining a knowledge problem, selecting the problem handling team should be based on the knowledge domains needed for analyzing the problem.

The selection can only be done on the basis of the mental idea of the problem of the person(s) who select(s) the team. Based on how he/she/they see the problem, their ideas about which domains and disciplines are involved, the domains and disciplines can be selected. In the beginning of the problem definition there is often only a very vague idea of the domains and disciplines involved in the problem. Before identifying the domains, the initiators can interview a number of experts in the field, or people who have an overview of the phenomena involved in the problem. In this way they can try to find out what domains are involved. Based on these interviews they can indicate the domains and disciplines that are needed. As indicated before, selecting a team already includes or excludes certain decisions.

Who should do the selection depends on who is responsible for the problem, or who takes the responsibility for the problem. The person(s) who is/are responsible for the problem handling should select the team or appoint somebody to select the team.

It is conceivable that more than fifteen experts are necessary, although the number of people forming a workable team would be too large. In this case one could select participants with knowledge of the major disciplines, and ask the other experts to join at moments where necessary.

The room in which the discussions are to take place should stimulate the information exchange by sitting in a circle, have room for separate meetings for small groups, add room for working alone. There should be sufficient technological support for group processes with telefax, telephones and other connections with people outside the building.

The participants should be able to meet each other regularly to discuss the problem. These meetings can be alternated with written information exchange. The written information exchange makes it possible for the participants to think matters over quietly, to gather new data, to consult colleagues and to summarize their knowledge about the problem for the other participants. Writing can also be performed at the place and the times that suit the participants.

8 The method should alternate individual preparation with group sessions. Although the emphasis is on group sessions where the participants can discuss with each other, the participants should also be able to have time for individual preparation of papers and lectures about their domain knowledge and take time to think and contemplate the problem.

9 The method should enhance domain support for the individual participants so that the participants can discuss with and get support from their own colleagues.

The team should be guided by a facilitator who is specially trained in applying the method and who has a neutral position towards the result of the discussion. The facilitator should be impartial towards the result and the participants. She or he should be an experienced, well-trained person in guiding group-processes and problem analysis.

Each person has knowledge of only a part of the problem. For defining the problem one should combine these partial knowledge sources into one overall view of the problem. Therefore the method should support the information exchange between the participants. This can be done partially by written text and partially verbally. In the verbal information exchange there should be room for discussions, explanations, lectures and asking questions in structured sessions and in informal meetings during and directly before or after the sessions. In this way the participants can share ideas, criticize and try out new ideas. In order to make this possible, the participants should meet each other regularly during the process of problem defining.

In a group of people it is quite normal to form coalitions. But we doubt whether forming coalitions would be fruitful in this case. The team spirit should not be interfered with by different parties, since this prevents people from stating their own opinion. Everybody should feel free at any moment to give his or her opinion on every subject. Therefore the method should not encourage bilateral contact between the participants or between the facilitator and a participant outside the group sessions.

three. In phase 1.3 the mental idea of the problem is extended into an iterative process with data fed by reading, talking, thinking or making observations about the problem. In this phase the problem, the concepts and the theory about the connections between the phenomena can be described in words. In an iterative process of connecting the mental idea with the data, the description of the problem can be progressively extended.

For phase 1.4 the method should support the formation of hypotheses about how the phenomena are connected with each other.

14 The method should support the process of definition on the basis of theoretical ideas.

The description of the phenomena involved and their relations are often based on a mix of assumptions, hypotheses, theories, experience and intuition. The method should strongly stimulate expressing on which ideas the description is based. What are assumptions, what are hypotheses and what is based on theory, experience and intuition? Where do these assumptions, hypotheses and theories, experiences and intuitions come from? Analyzing the problem further and gathering more information is the only way to strengthen or weaken these assumptions, hypotheses or theories, experiences and intuitions.

In an interdisciplinary discussion, the concepts that are used should be carefully defined in order to avoid misconceptions. Therefore the method should explicitly offer the opportunity to define the concepts and phenomena used.

16 The method should encourage the participants to confront with non supporting data[57].

One forms, as described in chapter three, a conceptual model of a problem in an iterative process by combining one's mental idea with data, and thinking and reading. We have seen that in this process one tends to look for supporting data and avoid data that is in contradiction with one's ideas. Therefore, the method should not only stimulate finding the supporting data, but should support explicit search for data that are in contradiction with the assumed ideas.

There is always a lack of knowledge and data in complex interdisciplinary societal problems. The method should stimulate one to fill these white spots.

18 The method should support discussing the aggregation level and the scope of the problem.

As discussed in chapter three, a complex interdisciplinary problem can have consequences at different aggregation levels, the micro, meso and macro levels. Not distinguishing these levels correctly can be the cause of misunderstanding which can be avoided by a careful definition of the level of the problem the discussion is about. In addition the scope defined for the problem handling process must be made clear to all the participants.

The language in which the model of the phenomena and the relation between them are expressed should be easy to understand for all the participants.

As indicated in chapter three, a model can be used as an intermediary between a theory and empirical data. First the problem can be described in words. This can be done by making a semantic model and in the case of a problem in which there are causal relations, a causal model. The method can assist in construction of a verbal causal description on which a causal network is based. These networks can be represented graphically. The graphic causal network illustrates how the phenomena are connected with each other and makes it easier to visualize the relation. On the basis of this description, the description of the phenomena involved can be combined into a (causal) network. The (causal) networks can be the start for making a system dynamic model. A problem that can be analyzed in terms of causes and effects should be expressed in a language that can express cause and effect. One of the languages that combines differential equations and cause and effect loops is the system dynamic language. A problem that is dynamic, that changes over time, can be expressed in differential equations. Time changes can be made visible with differential equations.

21 The method should stimulate an iterative way of going through the complete model.

Making one model within the seven layer model has consequences for the other models. The description of each model may make the former model more clear, or may result in an alteration of the former models. Therefore the modeling should be repeated so that the knowledge of learning effect of one model can have influence on the description of the others. With each description the problem becomes more clear.

By the core model we mean that first the macro aggregation level, then the specific scope of the problem should be viewed before going into more detail. At the moment the problem is defined, the connections with the whole macro aggregation level should be considered again[58].

23 The method should support, besides the modeling of the whole problem, the modeling of the domain models.

In order to understand the relations in the domains, domain knowledge should be modeled in the same way in which the whole problem is modeled.

As we have seen in this chapter there are several things one could do in order to prevent 'group think', although there is no guarantee that it can be totally avoided. As far as possible one should, and therefore the method should support[59]:

- forming parallel ad hoc (sub-)groups on crucial moments in the process and comparing the results of the discussion of these groups afterwards (3)

- frequently and continuously feeding back from the participants to experts in the same field (4)

- introducing people from outside to join the group for one or two sessions to give a critical review (5)

- inviting each person at a time deliberately to take the role of the devil's advocate (6)

- having special sessions in which one tries to find the weak points in the discussion (7)

- holding a special meeting, after a critical decision, in which everybody plays the role of the devil's advocate (8)

- periodically invite outsiders to join the group process for critical observation (9)

- having a parallel team, for very important problems to analyze and define the same problem at the same time.

Although we are aware that collective blind spots because of the complexity, can probably never totally be avoided, there are some things one can do that might help prevent these blind spots. The same steps taken to avoid 'group think' will, to a certain extent, also help to avoid collective blind spots.

In trying to avoid collective blind spots explicitly one can, at the end of each session, ask the question: "What have we forgotten?".

In order to make it understandable that there are such things as collective blind spots, one could present the knowledge graphically, as for example in knowledge islands.

26 The method should avoid, as far as possible, the negative influence of power differences among the participants.

In co-operative problem handling, one person can dominate another by difference in gender, professional status or character. But since everybody is an expert in his or her own field, the information and the contribution by each person is equally important. It is the task of the facilitator to see to it that everybody has equal chances to participate. This can be done by giving everybody equal time to speak, by letting each participant in turn comment, by sitting in a circle instead of a row. The facilitator cannot prevent power differences in every situation. In some cases power differences can be neutralized by guaranteeing anonymous answers. In this way nobody knows who has said something. This can be important in situations where critical comments are needed, in cases of the role of the devil's advocate, in brainstorming sessions and in voting situations. Special software has been developed for anonymous brainstorming and for anonymous voting and uttering criticism.

In this chapter we have indicated some of the conditions that should be fulfilled by a method for supporting complex interdisciplinary societal problem handling. We have done this based on the theoretical ideas of chapters two and three. On the basis of the discussion of chapters four, five and six, we have explored the way the computer can support the process of handling a complex interdisciplinary societal problem. How the method and the support of the computer are combined will be described in the next chapter.

for analyzing and defining complex interdisciplinary societal problems a method is needed that supports co-operative problem handling

The knowledge of the different aspects of the problem is (partly) available in the minds of human experts on different disciplines. In order to get a good insight in the problem, the personal knowledge of a part of the problem should be supplemented and combined with the views of other parts of the problem, held by experts of other professions. We saw that due to the complexity, the importance and the different disciplines, involved, the problem should be defined in co-operation. For this reason, defining these kinds of problem needs to be co-operatively undertaken by a multi-disciplinary team.

Each expert regards the problem from his or her own point of view, which is influenced by his or her profession, history, experience and political point of view. The process is interfered with by all kinds of negative group processes, like 'group think', hidden agendas and collective blind spots. In order to support the co-operative problem analysis of a complex interdisciplinary societal problem, a special approach is required that can be set out in a special method. In order to prevent the negative points of co-operative problem defining and in order to support the positive points, we listed twenty six items which a method for analyzing and defining complex interdisciplinary societal problems should support. We have thus answered the research question four:

what are the special aspects that a method for supporting the process of analyzing and defining complex interdisciplinary societal problems should possess?

[1] By 'handling a problem in co-operation' we mean handling a problem with a team. The arrangement of the group depends on the problem. It can vary from a mono-disciplinary team consisting of people of the same level of expertise, to a multi-disciplinary team of different (high and low) levels of expertise, or a mixed team with content experts and organization experts (managers).

[2] See for instance Marx' concept of false consciousness/awareness (Marx, 1890) and the ideas of Beauvoir (1949/1978) and Firestone (1970) about the position of women in society.

[3] Only secondary prevention includes curing.

[4] Secondary and tertiary prevention was seen during the flooding of the river Maas in the Netherlands between December 15 and 30, 1993, because the primary prevention proved to be inadequate (Journals Volkskrant and NRC December 15-30, 1993).

[5] See for instance the rules for kosher food preparation in the orthodox Jewish community.

[6] See for instance rules of behavior between people of a different caste in India.

[7] When the problem handling process focuses on tertiary prevention, there must also be attention given to primary and secondary prevention of the problem.

[8] An action or pressure group is a group of people who want to bring a problem under (political) attention. An action group is often an ad hoc group, that works at short notice; a pressure group, which often begins as an action group, is a more formal group, a more constant group which works toward long term goals (see also Rosenthal, 1984). A good description of a pressure group in the USA can be found in Shilts' book 'And the band played on' (1987).

[9] In the sixties and seventies there were many militant action groups such as the environmental activists, feminist action groups (Firestone, 1970; Boston Women Health Book Collective, 1978; Ekelschot & van Baalen, 1982; Grimbergen, Huibers & van der Peijll, 1983; Bussemaker, 1985; Faludi, 1991; Greenpeace).

[10] The car industry pressure group (BOVAG in the Netherlands, the American Automobile Association, AAA).

[11] Defining a problem as a mono-disciplinary problem does not automatically mean that these problems should not be handled by a team of persons. Many complex disciplinary problems can also be more fruitfully analyzed in co-operation than by one person alone. An example of this is analyzing the human genes, which is done by many researchers from many research institutes.

[12] The drug problem in the Netherlands is an example of policy making is that has been successively identified as, for instance, a legal problem, a medical problem and lately as a spare-time problem (Cohen, 1990). An example of a scientific approach to labeling a problem as domain specific, rather than interdisciplinary, is the opinion of many economists that technological development, as a result of economic development, can be described by only a few economic variables, like the costs of labor and capital (Leydesdorff & Besselaar, 1994). However, in our view the drug problem and the technological development process can only be analyzed and understood as an interdisciplinary societal problem, influenced by many phenomena and many people (DeTombe, lecture, 1993).

[13] Co-operative does not assume harmony.

[14] See chapter two.

[15] Most people are only educated in one field, at most in two.

[16] See early literature on women studies (Mies, 1978).

[17] On influencing each other's mental idea through conversation, see the conversation theory of Pask (1975). On influencing each other's concepts, see the research of Zelger (1993a).

[18] We prefer the term 'participants' to the term 'experts' because the term 'experts' has a connotation of a field/knowledge expert. By the term 'participants' we would like to indicate all sorts of people who are experts in their own field such as a manager (who is an expert in organization), a knowledge expert in a domain, for instance a professor, or a mother and housewife who is an expert in family life etc. The term 'expert' could easily suggest only people with an academic degree.

[19] Five to fifteen participants for instance, see discussion about the amount of people in chapter eight.

[20] People tend to value the information or the opinion of a person of a 'higher' profession more than the information or opinion given by a person of a lower ranked profession.

[21] Conclusions are given such as "23 people have this opinion and 5 people have that opinion".

[22] A facilitator is a person who guides the problem handling process. See also this chapter, section "The group must be guided".

[23] Research on the issue of group-think derives mostly from the analysis of decision processes that had a negative effect.

[24] Translation by the author.

[25] The facilitator or the initiator can have these as well. Sometimes the facilitator selects and guides and interprets data in such a way that the hidden goals of the initiator and/or facilitator will be met, such as, for instance, guaranteeing that other assignments of the initiator are given to the facilitator or that the group process will lead to certain political goals (initiator). The hidden agendas of the facilitator and/or initiator can interfere dangerously and conflict with the optimal handling of the problem. However, it interferes mainly with the way a facilitator should guide the process, namely neutrally, in the direction of the outcome of the process.

[26] See also for remarks about the devil's advocate (Flood, 1991, p. 127-128).

[27] The psychological effect of the devil's advocate can be contradictory or even worse than group thinking (from Rosenthal, 1984). Many powerful politicians cannot deal with negative messages, preferring to kill or dismiss the messenger. Many dictators like to surround themselves with good news messengers (see for instance Greek plays, Mao Tse Tun, (Chang, 1991); Stalin, (Rybakov, 1988).

[28] Blind spots should not be confused with white spots, see section 3.3.5.4.

[29] These are the same things that could serve to avoid 'group think'.

[30] When the group meetings are frequent it can take about two or three days a week, including preparation time and writing reports of the process. These data come from experiences at Delft University of Technology, in the period July 1993 - May 1994.

[31] As for instance, a consultancy agency.

[32] See urgent problem in section 3.2.2.

[33] The phase of awareness and forming a (vague) mental idea, phase 1.1.

[34] It may happen that some people have already formed ideas about interventions or that the initiator likes to begin directly with the second sub-cycle. However, it is advisable in almost all cases where problem handling of complex interdisciplinary societal problems are handled, to start with phases 1.2, the phase of extending the mental idea by hearing, thinking, reading, talking and asking questions about the problem.

[35] By the term 'theoretical ideas' we indicate assumptions, hypotheses, theories, experiences and intuition.

[36] A causal model is made when it is possible to express the relation between the phenomena in causal dependencies.

[37] For more details see chapter eight and appendix I.

[38] See section 3.3.5.

[39] Only as an intelligent front-end of a database.

[40] In chapter eight this is described in more detail.

[41]The software package Hypercard is a package which can support these activities. Hypercard operates on an Apple Macintosh computer or Spinnaker Plus that operates on a Dos machine.

[42] There are also computer programs that support cognitive mapping, for instance the package Cope developed by the University of Strathclyde.

[43] See also the discussion about system theory in section 3.4.4.

[44] Performed with Stella software (Apple).

[45] These packages are Stella, Ithink, for the Apple Macintosh computer, PowerSim for Dos computers.

[46] See also section 3.4.4.3.

[47] Author's note.

[48] See Appendix II.

[49] Osborn (1953) states that this way of working produces double the number of ideas. Stroebe (1992, 1994) wondered if it really was true that brainstorming in groups produces twice the ideas that could be collected from individuals. In his research he found that, together, individuals generate more and better ideas than they did in groups. According to Stroebe (1994), reasons for a negative group performance could be found in the waiting time before a person can utter an idea, combined with the short working memory of the human being. One of the reason people are so enthusiastic about brainstorming may be that many people have the idea that the good and creative ideas are theirs.

[50] Using brainstorming this way may take away many of the objection Stroebe has against group brainstorming.

[51] Empirical experience of the author as a facilitator by using the software Group System V at a group support room at Delft University of Technology in the period July 1993 - May 1994, held in order to support complex interdisciplinary societal problems. Group System V is groupware software developed by the University of Arizona, that can be considered as a special Group Decision Support System.

[52] Later studies might extend the description.

[53] A carefully designed empirical study which the author has planned for the next four years may give some more view on the expectations in this study.

[54] Although the method can be used for complex organizational problems as well.

[55] Outshoorn (1986) describes how to put something on the (political) agenda.

[56] See section 3.3.

[57] In order to avoid discussion over the distinction between data and information, we use data in situations where information can sometimes also be used.

[58] See section 3.3.7.

[59] The numbers in brackets refer to points for avoiding 'group think' given in section 7.3.