Introduction

Conversations between professionals and non-professionals occur daily in the medical and education fields. In medicine, it is desirable that patients take part in their treatment (Report to the Parliament Number 21 1998–99), and that doctors possess the communicative skills necessary to promote both partnership (European Academy of Teachers in General Practice 2005) and shared decision-making (Muller-Engelmann et al. 2008). Within schools, the mandatory parent-teacher meetings should be characterized by dialogue (Report to the Parliament Number 30 2003–4; Report to the Parliament Number 31 2007–8), good two-way communication (Ministry of Education and Research 2006), and agreement (Education Act 1998).

However, in medicine informed decision-making is often not implemented (Braddock et al. 1999; White et al. 2007), and dialogue in parent-teacher conferences is frequently absent (Nordahl 2000, 2003). Knowing that ‘decisions of street-level bureaucrats … effectively become the public policies they carry out’ (Lipsky 1980, preface) underlines the seriousness of this gap between ideal requirements and actual practice.

People with disabilities deal with complex challenges concerning their treatment, education and daily activities, so in professional conversations they need to arrive at well-thought out decisions based on their special needs. Parents of children with disabilities often experience not being listened to when giving their definition of the problem (Lundeby and Tøssebro 2008). It is therefore of great interest to reach a deeper understanding of how communicative patterns in such conversations can be characterized and influenced (Efraimsson et al. 2004). Furthermore, an exploration of how to implement ‘communicative democracy’ concerning people with disabilities is requested (Karlsson and Hydén 2007). For people with disabilities, the problems are rooted in an oppressive and discriminating society (Oliver in Söder 2009). Realizing deliberation in professional conversation is therefore one step towards counteracting discrimination, a factor that is especially important to this vulnerable group.

As outlined above, decisions in professional conversations should be based on dialogue, agreement, participation, shared decision-making, and partnership. In this article, it is assumed that this can be realized through deliberation. Deliberative conversations are characterized by having equal participants, no coercion, and a rational, argumentative procedure oriented towards consensus (Habermas 1996, 305) Thus, the procedure of how the relevant issues are addressed makes a difference (and not only the content).

Purpose of the study

The focus in this study is on whether it is possible to evaluate the procedure of professional conversations. We want to develop an instrument which compares the presence of ideal deliberative principles with what actually takes place in professional conversations. The normative assumption is that the output of such evaluation may serve as input to increase the quality of decision-making.

Previous research

There are several studies aimed at evaluating professional conversations. For example, Karlsson and Hydén (2007) have analysed democracy in the rehabilitation sector, focusing on the participants' quantitative dominance and topic control by measuring the number of words and turns taken in the conversation. Roter Interaction System (RIAS) (Roter 2006; Roter and Larson 2002), which is empirically based, measures doctor-patient interaction, especially socio-emotional exchanges. Braddock et al. (1999) use content analysis to reveal the presence of informed decision-making between patient and physician. In political science, we find studies more strictly focusing on measuring deliberation procedures, especially evaluating online discussions (Graham and Witschge 2003; Hagemann 2002; Trénel 2004) and parliamentary debates (Steenbergen et al. 2003; Steiner 2004).

No instruments today can be directly used to evaluate whether principles of deliberation are present or not in professional conversations.

The development of the instrument

The theoretical framework is Habermas’ (1984, 1987, 1996) definition of deliberation. The composition and components of the instrument are mainly derived from Discourse Quality Index (DQI), a tool based on Habermas's discourse ethics and used to analyse deliberation in parliamentary speeches (Steenbergen et al. 2003). Preliminary editions of the instrument were adapted for practice by two coders discussing the coding of four transcribed conversations (three within the educational field and one within medicine) until agreement was reached.

Theoretical framework

According to Habermas (1984, 1987, 1996), in a deliberative dialogue all parties concerned must have the equal right to participate in an equal manner without use of internal or external coercion. The dialogue has to be oriented towards achieving mutual understanding being based on rationality. Participants are obliged to give reasons for any statements made by them if this is requested by someone else. In this way, the procedure takes on an argumentative form through the regulated exchange of information and reasons. The taking of yes/no positions should be motivated solely by the force of the better argument (Habermas 1996, 305). The participants should be open to change their minds when convinced, and their statements must meet validity claims of being true, right and truthful. The aim of deliberation is to reach consensus based on the same reasons. Consensus is the result of deliberation and not considered part of the deliberation procedure, and is thus not included in the definition of deliberation.

The composition of the instrument

We have used the composition of DQI as a template, with adaptations for professional conversations. The conversations to be evaluated have been recorded and transcribed verbatim, including indications of pauses, laughter, ‘hmm's’, etc., catching the auditive elements and excluding visual elements according to the nature of transcribed material (Mishler 1984). The transcription has been set up for direct coding in Excel to facilitate the computing of the data material (see the example in Appendix 1). The units of analysis are statements: from the moment the participant starts to speak until their mouth closes once again. Each unit of analysis has been assigned one cell in the spreadsheet. In the same row, variables and their respective indicators have been presented and scored according to the presence or absence of the variables in that actual statement. Presence is the same as ‘yes’ and is given code 1, absence is ‘no’ and coded 0. The variables and their indicators are defined in the following section. All statements are considered relevant. Each unit of analysis has been labelled with the participant to whom it belongs. The discussed theme(s), participants and topics have been assigned numbers in their process of emerging. When someone remains silent despite having been given the opportunity to speak, the silence is still considered a unit of analysis and thus coded. If there is any doubt as to how a statement should be comprehended, the coder has interpreted the meaning in accordance with what he thinks is favourable to the person making the statement. If a statement is repeated, it is valued each time, as the participant may have an understanding that the other participants did not fully comprehend when the statement was made for the first time.

Each contribution to the conversation is considered as (potentially) both context-determined and context-determining, context meaning the local dialogue context (Linell, Gustavsson, and Juvonen 1988). Hence, in the coding process all statements have been compared both to the previous statement and the conversation as a whole. We also find it important to interpret the results according to what could possibly be expected of the participants, bringing in context in a wider sense. This demands for ‘fronesis’, that is, deciding what ‘good’ practice is depending on the situation, accessible means, the person(s) involved and the progress of the process (Aristoteles 350 BCE/1934).

The name of the instrument is Measuring Communicative Deliberation (MCD), inspired by the DQI and the concept ‘communicative deliberation’, the latter developed within the education field (Englund 2006). Table 1 offers an overview of MCD. The first column contains the variables included in the instrument presented (numbered from 1 to 8). The name of the variable is outlined in column two. Each variable is operationalized, and their indicators are described in column three along with their corresponding codes. In the next section, the operationalization process is described.

Table 1. Presentation of MCD with variable numbers, variable names, and the indicators by which each statement is to be judged, indicating whether the variable is present or not.

  Variable indicator
Variable number and name No Yes
1  Respect for possible  participation The statement is an interruption (code 0) The statement is not an interruption (code 1)
2  Implicit respect for a person No respect: This code is reserved for speeches in which there are only negative statements towards a participant, opinions or values. The code is also used if the speaker talks in a disrespectful manner (code 0) Implicit respect: We use this code if there are no explicitly negative statements, nor any explicitly positive statements and matters directed towards a person, opinions or values (code 1)
3  Explicit respect for a person No respect (code 0) Explicit respect: This code is assigned if there is at least one explicitly positive statement about participants or matters concerning participants, opinions or values, regardless of the presence of negative statements. Participants are defined as both the person spoken to and as other persons present in the room (code 1)
4  Disrespecting a person One of the variables implicit/explicit respect is found (code 1) Neither of the variables implicit/explicit respect is found (code 0)
5  Respect for last statement The last participant's statement is ignored (code 0) The last participant's statement is included and either degraded or valued. This code is provided even when the statement is an interruption (code 1)
6a Promoting mutual understanding once Informal talk or silence (code 0) The statement promotes mutual understanding that is relevant information, relevant opinion, relevant argumentation, a relevant question, or talk about the conversation (code 1)
6b Promoting mutual understanding more than once Informal talk or silence (code 0) Either two of the same elements or more than one element defined as promoting mutual understanding (code 1)
7a Expressing understanding or agreement once The statement does not contain understanding or agreement of the other participant(s) (code 0) The statement contains one understanding or agreement of the other participant(s), including rephrasing of statements and/or establishing a joint understanding of arguments (code 1)Helpful sentences: Yes, that's the way….Yes, that's right…Yes, that's how they see it tooYes, that's the way I see the world tooYes, then we agree on…. Then we do this….Yes, I agree to do thisI agree to valuate this argument…
7b Expressing understanding or agreement more than once The statement does not contain understanding or agreement of the other participant(s) (code 0) The statement contains two or more understandings or agreements (code 1)
8  Expressing disagreement The statement is not an explicit disagreement (code 0) The statement expresses a disagreement (code 1)

Operationalization of deliberative principles into variables

Derived from the theoretical framework, the deliberation principles to be operationalized into variables are: ‘Equal right to participate’, ‘equality’, ‘no use of coercion’, ‘mutual understanding’, ‘open to changing one's mind’ when convinced, and ‘fulfilment of the validity claims’. Below is a step-by-step description of the development of the operationalization into variables and their indicators.

The variables have been illustrated through examples taken from two transcribed professional conversations. One takes place in school between a girl with learning disabilities, her mother, the class teacher, a pedagogical-psychological counsellor and a special needs teacher. The purpose of the meeting is to discuss which course of study the girl is to choose in upper-secondary school, including necessary adaptations. The other conversation used as an example is a doctor-patient consultation where the subject is following up a patient who needs to lose weight.

1. ‘Equal right to participate’, ‘equality’, ‘no use of coercion’

Power unfolds and is exercised in the complex practices and relationships between teacher-pupil and doctor-patient (Vågan and Grimen 2008). In almost all interactions, both parties have some power (Linell, Gustavsson, and Juvonen 1988, 437). Hence, power abuse in an asymmetrical relationship is not always related to the professional. A pupil or patient may use power, for example, by means of his/her resistance to talk. Additionally, coercion might take place in a multitude of other ways; through the use of words, body language, the placing of furniture, the design of the room, the manner of deciding the agenda, and so on. According to this, power in our study is seen ‘in terms of underlying (multidimensional) structures regulating interpersonal relations or as POTENTIALS to influence other people's behavior and thinking, coupled with beliefs and expectations (on the part of others) that these potentials may be used’ (Linell, Gustavsson, and Juvonen 1988, 437). This displays that the concepts ‘no use of coercion’ and ‘equality’ contain complexities which are impossible to fully uncover and measure in real life, and certainly not in transcribed text. To capture at least some of these aspects in a text, they need to be broken down into more concrete and multifaceted elements. After looking more thoroughly at the other deliberative principles, we realize that the breach of every single one of them is also a breach of the elements of ‘equality’ and ‘no use of coercion’: To exclude someone from ‘participation’, either from verbal participation by never being allowed to speak or excluded from being present at all, is clearly power abuse. This may also be the case if the words or language are unfamiliar to some participants, if all information is not shared or if all participants' opinions and viewpoints are not asked for, breaching the principle of searching for ‘mutual understanding’. Moreover, not ‘fulfilling the validity claims’ by, for example, not telling the truth, is also an abuse of power, as is also the attitude of not being ‘open to changing one's mind’. Consequently, operationalization of the concepts ‘no use of coercion’ and ‘equality’ may partly be reflected in all the other deliberative principles. The concepts are also reflected in the term ‘respect’, used in DQI and found to be an element of deliberation in several studies (Ackerman and Fishkin 2002; Greenhalgh, Robb, and Scambler. 2006; Steenbergen et al. 2003). In MCD, we connect the respect variables to the participant's behaviour, emphasizing that each participant (as well as her or his values and choices) must be met with respect (Eekelaar 2006) in order to create an atmosphere of freedom in which information and opinions can be presented.

In DQI, ‘participation’ displays the possibility to take part in a discussion without being interrupted. In our instrument, the possibility to participate is understood as expressing aspects of respectful behaviour, and our variable is called Variable 1, Respect for possible participation. The indicators for this variable are maintained from the DQI. An example of coding Variable 1, Respect for possible participation, is taken from the conversation in the school context:

Pedagogical-psychological counsellor: …What do you think is important to consider when applying [for school]? I think that what's important for S – is not to drop that possibility but to look at…

Mother: I also think that's a really good suggestion.

This is interpreted as an interruption and is coded as 0. When a participant clearly completes the last speaker's statement, we do not interpret this as an interruption. As illustrated by Bråten (2007), sentence completion can be an indication of ‘moment of meeting’ (Stern 2004), an intersubjective sharing of common experience. When a patient or a pupil does not answer when asked, the subsequent silence which is interrupted by another participant may construct the patient/pupil as a half-member of the interaction (Hutchby and O'Reilly 2010), or it may be interpreted as caring for that person. We have chosen the last interpretation and do not code this situation as an interruption. With respect to ‘participation’ in the sense of being present at all, MCD presupposes that all concerned are present, even if this is not always the case in professional conversations.

DQI has a variable called ‘respect for groups’. To adapt this respect variable to professional conversations, we replace it with Variable 2, Implicit respect for a person, Variable 3, Explicit respect for a person, and Variable 4, Disrespecting a person. The coding indicators are, however, the same as in DQI.

An illustration of Variable 2, Implicit respect for a person, is taken from the doctor-patient conversation. The doctor asks the patient why he wants to discuss lifestyle changes:

Mmm, can you, Patient 1, say anything about why you, in a way, have now come to a point … where you want to lose weight and almost … do something with your life so that you…

As this statement contains no explicitly negative or explicitly positive statements, it has been assigned Code 1.

An example of Variable 3, Explicit respect for a person, is from the school context when the special needs coordinator states that she enjoys having the pupil at school:

Actually, we didn't think that you would be staying here beyond the summer, even if it is very nice to have you here. Now you're going to go on to upper secondary school.

This statement contains one explicitly positive statement about another participant, Code 1.

From the same conversation we identified the following example of Variable 4, Disrespecting a person, when the mother states that her child is sometimes mean:

Sometimes in some things she is, well she's – basically very, very difficult – sometimes she's very mean, so that I have to grab her by the scruff of the neck. I think she was nasty when she grabbed her brother's nose and twisted it so it turned black and blue.

Neither the variable Implicit respect for a person nor the variable Explicit respect for a person, is found in this statement, and it is therefore encoded 1 for Variable 4. Variable 4, Disrespecting a person, is actually redundant, as its outcome is a pure combination of Variable 2, Implicit respect for a person and Variable 3, Explicit respect for a person. Despite this redundancy, we chose to include Variable 4 to emphasize the occurrence of showing disrespect. Note that Variable 4 is formulated as a negation, and the translation of yes/no into numbers in the spreadsheet must take this into account. In the statistical analysis, all other ‘yes’ codes are translated to the value of 1 while no (or lack of yes) is translated to 0. Regarding Variable 4, Disrespecting a person, this is the other way round (yes encoded as 0).

DQI includes the variable ‘respect for counterarguments’, meaning that a speaker includes and evaluates earlier counterarguments in his or her statements. In conversations, people strive for acknowledgement of their communicative input, and consider natural response and follow-up as a confirmation and a non-response as de-confirmation of the speaker (Marková 1991, 236). Hence, the DQI variable is made relevant for conversations by valuing a natural follow-up of the last participant's statement, in MCD called Variable 5, Respect for last statement, with indicators as shown in Table 1a. The example of Variable 5, Respect for last statement, is a sequence from the patient-doctor conversation where neither the doctor nor the patient shows respect for the other's last statement. The patient tries to explain how he tries not to eat at work while the doctor tries to explain that the patient needs to take a blood test:

Patient: … try not to take money with me to work …

Doctor: Then we're going to…

Patient: … so I don't buy anything

Doctor: … take that … test now, and then…

The doctor and the patient totally overlook each other's statements, reflected by Code 0 for Variable 5 for both. A similar variable is also used by studies evaluating deliberation in Internet debates, focusing on the possible power abuse embedded in ignoring statements (Dahlberg 2001; Graham and Witschge 2003; Hagemann 2002). If the conversation is characterized by statement which does not follow up the last speaker's statement, this might be a sign of monological dialogue where the speaker regards his statement as exhaustive with no need for additional comments (Seikkula 1996). Monological expressions may also be formed as questions, where the goal of the questions is to search for clarifications of one's own assumptions. Such expressions are to a certain degree necessary, for example, during parts of diagnostics in medical consultations. They promote understanding, but not necessarily mutual understanding. To fully reveal the presence of monological expressions, we suggest qualitative analysis of the text.

2. ‘Mutual understanding’

DQI operationalizes ‘mutual understanding’ by valuing the presence of justifications in statements. They also value the content of a justification if the speaker refers to group interests or appeals to the common good. Referral to the common good is relevant in professional conversations, for example, when the professional refers to the law or tight finances. For the most part, however, the participants discuss more private topics. To simplify the instrument and focus on the most relevant parts, the valuation of group interests and the common good are omitted in the MCD; however, they retain their value if justification is present or not. Justification is called ‘relevant argumentation’, similar to Trenel's wording (2004). In accordance with studies measuring deliberation on the Internet, MCD also includes other elements found to promote mutual understanding: that participants provide and ask for information (Dahlberg 2001; Hagemann 2002; Muhlberger 2006), that opinion interchange is present (Hagemann 2002; Muhlberger 2006), that the participants talk about the conversation, that is, meta-talk (Muhlberger 2006), and that agreement and disagreement are pronounced. (Muhlberger 2006). Concerning the last point, pronounced agreement can visualize opportunities for establishing consensus, and pronounced disagreement can reveal the need for argumentation or further exploration of the topic. This may be easy to forget and is not always necessary in everyday talk, while it may be of crucial importance in professional conversations to increase transparency concerning whether common ground has been found or not. Correspondingly, it appears important to actively express that a statement has been understood, which is also considered part of deliberation (Muhlberger 2006).

All in all, it seems desirable to incorporate the following elements in the deliberation principle ‘mutual understanding’: relevant argumentation, giving and asking for information, opinion interchange, meta-talk about the conversation, pronounced agreement, pronounced understanding of another person and declared disagreement.

In earlier versions of the MCD, we tried to separate these ‘mutual understanding’ elements. This turned out to be difficult, as the two coders repeatedly coded utterances dissimilarly. As the main objective of the instrument is to evaluate contributions to mutual understanding and not necessarily to separate each element, the pragmatic solution to this challenge was to merge the first four (out of a total of six) elements of ‘mutual understanding’ (i.e. argumentation, giving and asking for information, interchange of opinions and meta-talk about the conversation) into the Variables 6a, Promoting mutual understanding, and 6b, Promoting mutual understanding more than once. An example of Variable 6 has been taken from the conversation at school. The special needs coordinator asks if the pupil sometimes thinks about what her life will be like after she has completed school:

Special needs coordinator: Do you sometimes dream about what you will do when you've finished school? What do you dream about how you'll be earning money?

Pupil: Umm … working in a zoo or something?

This statement promotes mutual understanding through the asking of a question and has been awarded Code 1. The pupil has also been awarded Code 1 on this variable because her clear answer contributes to mutual understanding, and thus deliberation.

Informal talk may support deliberation (Habermas 1996, 308). The instrument makes it possible to unmask informal talk by looking for utterances coded 0 in Variables 6a and b.

The element ‘asking for information’ needs further comment: Linnell, Gustavsson, and Juvonen (1988) have developed the initiative-response instrument (I-R) which focuses on whether the statement is an initiative or a response to a statement; and where asking questions is seen as a way in which the parties try to force the interlocutors to respond on their terms. Professionals generally ask questions where clients answer without being able to see the wider context of the questions, thereby creating an asymmetrical conversation where the professional and client do not have the same aims, knowledge and resources (Efraimsson et al. 2006). In MCD, asking questions is considered a necessary part of the deliberations; nonetheless, the instrument does not capture whether the questions constitute the grounds for one of the interlocutors to hold model power (Bråten 1998). It is possible, however, to reveal this by identifying the statements scored as Variable 6 and undertake a qualitative interpretation.

The two remaining elements of ‘mutual understanding’, pronounced agreement and pronounced understanding, turned out to be difficult to separate as well; sometimes it seemed that pronounced understanding also included an agreement and sometimes not, and at other moments it was impossible to know whether the speaker agreed or not. Habermas outlines the distinction between agreement (Einverstandnis) and understanding (sich verstandingen) in the following way:

Agreement in the strict sense is achieved only if the participants are able to accept a validity claim for the same reason, while mutual understanding (Verständigung) can also come about when one participant sees that the other, in light of his preferences, has good reasons in the given circumstances for her declared intention – that is, reasons that are good for her – without having to make these reasons his own in light of his preferences (Habermas 1998, 321).

As reasoning can often only be decided by asking the speaker directly, this explains our difficulties in separating understanding and agreement, when this is not pronounced explicitly. Based on this background, we found it necessary to merge these two elements into Variable 7a, Expressing understanding or agreement once, and 7b, Expressing understanding or agreement more than once. As the intention of the MCD is to measure whether the process in the conversation is characterized by deliberative principles (and not a characterization of the degree to which a consensus is reached), this merging of the two elements has been found acceptable.

An example of Variable 7 has been taken from the doctor-patient consultation. The doctor and the patient first elaborate on the patient's medical condition, and then the doctor expresses understanding or agreement with the patient:

Patient: There is somewhat high cholesterol in the family. I know this.

Doctor: Mmmmmmm … inherited problem here.

The doctor's statement contains one explicit agreement with another participant and has been assigned Code 1.

In further testing of the MCD, it was also at times difficult to distinguish between Variables 6 and 7 (Promoting mutual understanding and Expressing understanding or agreement). Therefore, helpful phrases (presented in Table 1) were used to assist the interpretation. The idea behind this derives from Habermas (1984), who developed helpful phrases to facilitate the interpretation of statements. In MCD, if the original sentence can be replaced by one of the helpful phrases in Table 1, it is scored according to Variables 7a or 7b (Expressing understanding or agreement or Expressing understanding or agreement more than once).

In the following example of Variable 7, from the school context, the special needs coordinator emphasizes both understanding and agreement in her answer to the mother. Here they are discussing about how the girl should spend her money, a sub-topic in the conversation. Note that the coordinator's answer can be replaced by the helpful sentence ‘I agree to valuate this argument.’ This replacement might support the coding process.

Mother: It's very difficult for me to refuse to let her have her own money, but because she doesn't have any financial good sense, you know, I'm forced to…

Special needs coordinator: You have to.

The special needs coordinator's statement contains an explicit agreement and also shows understanding, Code 1.

The last deliberation principle of ‘mutual understanding’ is disagreement. In an earlier version of the instrument, we differentiated between demurring and explicit disagreement, considering demurring as a weak disagreement. The empirical testing revealed that demurring can also be understood as an argument and/or information. Explicit disagreement was withheld and demurring excluded, leading to Variable 8, Expressing disagreement. We do not provide an example of this variable, as none of the coded conversations in our empirical material contained explicit disagreement.

The MCD captures whether the elements that have been found to promote mutual understanding are present or not, and how the total number of statements has been distributed between the participants. However, there is no necessary or general connection between interactional control and amount of speech (Linell, Gustavsson, and Juvonen 1988, 436). In our example, the girl with learning disabilities scored relatively high on ‘mutual understanding’. While we do not know whether the score represents initiatives or responses, this can be revealed by a qualitative study of the girl's score, which might offer important additional knowledge about the girl's role in the conversation. As argued by Marková (1991), 221), in referring to Leudar, people with disabilities are often placed in a non-reversible role and are not given equal opportunities to initiate and perpetuate discourse, question anything, make statements or express their attitudes and feelings openly.

It is also important to note that the Variables 7a, Expressing understanding or agreement, 7b, Expressing understanding or agreement more than once, and 8, Disagreement, have always been additionally coded as information, that is, given points according to Variables 6a, promoting understanding once, and 6b, Promoting mutual understanding more than once, thereby creating an overlap of coding points.

3. ‘Open to changing one's mind’

The deliberation principle ‘open to changing one's mind’ is not directly assessed in the DQI, which has one related variable that has been called ‘constructive politics’, valuating orators who offer alternative propositions; however, this is not included in the MCD. According to Trenel (2004), being ‘open to changing one's mind’ may be expressed through empathy and respect, although there are no direct measurements of this deliberation principle in earlier research studies. The concept can be regarded as an attitude which is impossible to observe in a transcribed conversation. Although being partly reflected in our respect variables, this attitude cannot be fully revealed without directly asking the participants.

4. ‘Fulfilling validity claims’

One premise for a conversation based on deliberation is that the validity claims of truth, right and truthfulness are fulfilled. However, this is omitted in several studies measuring deliberation (Graham and Witschge 2003; Hagemann 2002; Steenbergen et al. 2003; Trénel 2004). Truthfulness, also called authenticity, is difficult to measure in texts. According to Habermas, lack of consistency between a statement and action is a sign of untruth (1984, 41), social pathologies are unmasked by contradictions (1987, 378), and lack of continuity and coherence are signs of a legitimation crisis (1987, 140). Bearing this in mind, consistency/inconsistency was included as a variable in an earlier version of the instrument. However, testing proved that inconsistency was never uncovered, perhaps because it was not present, or due to shortcomings in the instrument. Qualitative analysis is found to be a better method for discovering unfulfilled validity claims, these being inconsistency and lack of coherence.

An overview of the operationalization of the deliberation elements is given in Table 2.

Table 2. Operationalization of the main elements of the deliberation procedure into variables.

Variable Number Characteristics of Deliberation Operationalization
1–5 Equal right to participate MCD presuppose the presence of relevant participants.
  Equality and no use of coercion Reflected in all characteristics of deliberation, but also in the degree of respect.Measured by the following variables:Variable 1 Respect for possible participationVariable 2 Implicit respect for a personVariable 3 Explicit respect for a personVariable 4 Disrespecting a personVariable 5 Respect for last statement
6–8 Mutual understanding Measured by the following variables:Variable 6a Promoting mutual understanding once Variable 6b Promoting mutual understanding more than onceVariable 7a Expressing understanding or agreement onceVariable 7b Expressing understanding or agreement more than onceVariable 8 Expressing disagreementIn addition: Equality and respect creates an atmosphere promoting mutual understanding.
  Open to changing one's mind Partly reflected in respect variables. Cannot be fully revealed without asking the participants.
  Fulfilling of validity claim Reflected in inconsistency and lack of coherence, better revealed by qualitative methods.

The instrument's reliability

To test the inter-coder reliability of the instrument, a new conversation (from the school setting) was coded by the two researchers independently and then compared. Excel and SPSS were used as tools in this analysis. Each of the 138 units of analysis were evaluated according to 10 dichotomous variables (yes/no) to identify whether deliberation principles were present or not, producing a total of 1380 evaluations. Dichotomous variables were chosen because this has been found to simplify the coding process. A summary of the inter-coder reliability is shown in Table 3 below.

Table 3. Overview of inter-coder-reliability for each variable measured by point-by-point and Cohen's kappa.

Variable Point-by-point Cohen's kappa
1 Respect for possible participation 99% 0.74
2 Implicit respect for a person 91% 0.69
3 Explicit respect for a person 93% 0.73
4 Disrespecting a person 98% 0.56
5 Respect for last statement 99% 0.85
6a Promoting mutual understanding once 90% 0.65
6b Promoting mutual understanding more than once 93% 0.87
7a Expressing understanding or agreement once 86% 0.65
7b Expressing understanding or agreement more than once 95% −0.01
8 Expressing disagreement 100%  
All variables 94% 0.89

As illustrated, the point-to-point agreement varied between 86% and 100% for the 10 variables with an average of 94% (i.e. the coders encoded 94% for all 1380 evaluations identically). Thus, all the variables were within the range of a traditional guideline which, according to Kazdin (1982), is agreement at or above 80%.

To investigate if the agreement between the coders could be accidental, Cohen's kappa (1961) was calculated. This method takes into consideration the frequency of both occurrence and non-occurrence of the scores. Ciccetti (1984) has classified kappa values in the range of 0.40–0.59 as fair, 0.60–0.74 as good and 0.75–1.00 as excellent agreement. Regarding eight of the variables, Cohen's kappa was in the 0.56–0.87 range. The coding of these variables is therefore not found to be influenced by chance. Variable 8, expressing disagreement, has no occurrences by any coder and could therefore not be used as a basis for reasonable correlation calculations. Variable 7b, expressing understanding or agreement more than once, had a very low frequency, only three occurrences. Very few discrepancies between the coders will then invalidate the result. This is reflected by a Cohen's kappa at -0.01. It cannot be concluded whether this is due to the low frequency or if the instrument fails to capture these phenomena. The two latter variables should be developed more or omitted all together.

Discussion, conclusion and practical implications

Discussion

The validity of the instrument is found to be ensured by its theoretical grounding, reference to other measurements and reference to empirical material. Nonetheless, there are limitations to consider when the instrument is used. As made evident in the operationalization, MCD cannot fully capture whether deliberation is present or not. Further research is needed to disclose whether it is possible to reveal more of the deliberative aspects in transcribed texts.

The high inter-coder reliability is considered to indicate that it is possible to identify and understand aspects of deliberation in an unambiguous way in transcribed text. The high reliability of most of the variables indicates that the instructions for the instrument are acceptable. The variables disrespecting a person and, especially, expressing understanding or agreement more than once are characterized by low reliability scores, measured both by point-by-point and with Cohen's Kappa. These variables obviously need improvement. Concerning the variable disagreement, there is no expressed disagreement in the empirical examples. Of course, in some conversations absence of disagreement is acceptable. But this may also reflect a conversation with invisible opinions and, consequently, little mutual understanding. Thus, even though there are no manifestations of disagreement in our empirical material, the variable is withheld.

Even if the validity and the reliability of the instrument are found to be satisfactory, there might still be objections as to whether it is possible to use an instrument to identify deliberation principles, assign them a number, summarize the numbers and achieve a meaningful result. According to the authors’ experience – one a general practitioner and the other a former special needs teacher as well as pedagogical and psychological counsellor – the instrument also provides), in its present form, meaningful information on professional conversations.

It is also important to ascertain the extent to which the variables’ numerical values are in reasonable proportion to each other. Future research might reveal more correct proportions. Further development of the instrument may also fulfil the strict requirements for indices, thus creating a ‘mutual understanding index’, ‘respect index’ and overall ‘deliberation index’.

The final scores of the MCD focus on the conversation as a whole, even though the interactional pattern may vary in the various phases of the conversation. An improvement of the MCD may reveal such differences. Moreover, ideal values of the scores are not available. This, too, might be subject to further investigation. On the other hand, perhaps there are no ideal values. The instrument is in a way ‘tossed’ over the conversation like a fisherman's net, and it is important to be aware that in different conversations and for various participants the ideal sizes of the elements in the net are different. As mentioned above, several considerations have to be made when interpreting the scores and the interpreter of the results must exercise ‘fronesis’. For example, the expectations of a pupil with learning disabilities and a teacher should be different.

In MCD the moderator is treated in the same way as the other participants. A further development of the instrument might be to develop different variables for the moderator and other participants. However, in this version of the MCD as well, the variables relating to the moderator might provide valuable information as to how the moderator behaves, to guide the moderator in how to fulfil deliberation requirements in addition to providing empirical data as the starting point for reflection. For example, an active moderator might obtain a high score. While this might be positive, a moderator who asks a lot of questions might also impede the authentic voices of a patient or a parent/pupil and exercise model-power (Bråten 1998). The conversation might then be conducted on the premises of the professional. On the other hand, ‘allow to flow’ might have an important function (McWhinney 1997). The therapeutic role described by Anderson (2003) is also a succinct description of the moderator role in deliberative conversations, emphasizing that the moderator should avoid being the ‘narrative editor’: the client should tell his or her story, while the moderator's role is not to correct or impact the client's story. The moderator should also repress prior knowledge, focus on the expertise of the client and be a participant in the conversation. However, if a participant does not state his or her opinion, it might be the moderator's task to ask for it.

Habermas's theory is used analytically, even though choosing a theory is always a normative decision. The assumption in Habermas's theory is that deliberation is well suited for reaching conclusions which are considered good for all the participants. This idea is supported by White (1998), who claims that ‘deliberation may produce thoroughly reasoned decisions’. Shared decision-making is associated with greater patient satisfaction (White et al. 2007), suggesting a coherence between deliberation and patient satisfaction. This is in line with Chambers (1996, 2003), who argues that deliberation is suitable in one-to-one debates, and Emanuel and Emanuel (1992), who argue that a deliberative model is the preferred approach in a doctor-patient relationship. The deliberative principles can potentially broaden perspectives, promote tolerance and understanding and direct the focus on the communicative processes of opinion and will formation (Chambers 1996). The concept takes into account that the participants in a dialogue are not always sure of their own opinions (Emanuel and Emanuel 1992) and that there is a need for reflection between the doctor and patient (Walseth, Abildsnes, and Schei 2010). The reflection and search for shared values through justification inherent in deliberation constitutes a teaching process for the participants (Wayne 2003). Moreover, the deliberation process might ideally legitimate the decisions made (Habermas 1996, chap. 7). Consequently, the potential and reasonable consensus building inherent in deliberative theory appears to be suitable for professional conversations, and the procedure of deliberation may function as a guideline for the professional in his/her role as moderator. Used wisely, guidelines based on deliberative principles are considered well suited to taking care of the complexity of daily life for persons with disabilities and increasing the opportunities for their voices to be heard.

Habermas (1996) develops the idea of deliberation within the field of politics. Applying this in professional conversations, several issues have to be taken into consideration, especially relating to the fact that the ideal of deliberation implies equal participants (Walseth and Schei 2010). Professionals and non-professionals traditionally have an asymmetric relationship characterized by inequality. Asymmetry is considered an intrinsic feature of dialogue referring to inequalities concerning various background conditions, distribution of knowledge, and differences in social position (Linell and Luckmann 1991). One key challenge is that deliberative theory assumes communicative and cognitive competence (Barnes 2002), while professional conversations might be asymmetric in available information and personal abilities. The asymmetry of medical encounters is considered both a resource and a challenge, demanding patient-centred leadership characterized by empathy and the ability to assume the patient's perspective (Schei 2006). The vulnerability of a patient or a pupil and parent has to be taken seriously. The doctor has to offer empathy and be aware of vulnerability, assuming a leading role and using practical wisdom to adapt the consultation to the specific situation (Schei 2006; Walseth and Schei 2010). Additionally, there is a growing tendency for non-professionals to search for pertinent information on the Internet, which may be inconsistent with or go beyond the competence horizon of the professional. This new situation might challenge the traditional role of the professionals and non-professionals. However, ‘good conversations … presuppose academic humility and acknowledgment of scientific uncertainty, ignorance and fallibility’ (Walseth and Schei 2010), and demand that the professional is ‘open to changing his/her mind’ when convinced. Another challenge is the structural dependence inherent in the reliance on laws, rules, and how the actor complies with these in practice. A study of texts written by the parents’ organization reveals how the parents themselves describe their own role both as equal and dependent on the teacher, and by so doing, they may have pointed out a core challenge in the non-professional role (Tveit 2009b).

The degree to which the deliberation procedure should be realized supposedly varies during the different phases of a conversation and alters between conversations. In some conversations it is legitimate for the professional not to take deliberation principles into account, for example, in emergency situations. Research also indicates that teachers and parents do not hold the dialogue as described in the theory of communicative action as an exclusive ideal for their conversations (Tveit 2009a Tveit 2010); sometimes they, well reasoned, choose not to fulfil the validity claim truthfulness by prioritizing the concept of tact. Furthermore, it may not always be crucial to reach ideal consensus, as deliberation may still contribute to the task of reaching an understanding of each other's viewpoints, of great value in dialogues (Habermas 1984) and emphasized in consultations (McWhinney 1997). Moreover, as available time and resources are generally limited, the bureaucracy will limit the possibilities for the participants to reach ‘free’ decisions.

These circumstances perhaps make ideal consensus seem illusionary, and, as Habermas (1996, 326) argues ‘this model is merely a methodological fiction intended to display the unavoidable inertial features of societal complexity’. According to Pellizzoni (2001), it may still be useful to strive for Habermas’ idealized theory. Consequently, professionals should possess knowledge regarding deliberative procedures and, in adapting to the situation, deliberately choose whether or not to follow them. With necessary adaptations, deliberation is considered worth striving for, in many cases constituting an important condition for high-quality professional conversations.

Conclusion

Although in need of further methodological development and refinement, the present instrument was found to be an acceptable first edition capable of identifying deliberative principles in professional conversations. All in all, one might say that the MCD can provide information on how much each participant contributes and how each participant behaves. Such information is vital when the ideal of professional conversations is user participation and dialogue. The instrument might reveal information which can enhance the quality assurance of professional conversations and also be used in teaching at the university level.

Practical implications

This study has argued that following deliberative principles in professional conversations promotes good decisions. The results from the MCD might help make the participants aware of their behaviour and able to reflect-on-action (Schön 1995). By using the MCD as a basis for learning processes with the aim of increasing the quality of professional conversations, the participants may also be able to reflect-in-action (Schön 1995). When the instrument is applied in transcribed professional conversations, several meaningful aspects related to the presence of deliberation principles in professional conversations might be revealed. Possible outcomes might, for example, be the following:

  • •  A summary of the participants' behaviour with respect to equality and use of coercion
  • •  A review of the participants' contribution to achieving mutual understanding and agreement
  • •  A comparison of the participants' contribution; between professionals and non-professionals, different roles and between the genders
  • •  A characterization of the entire conversation
  • •  An overview of how each topic is treated
  • •  A comparison of the presence of deliberation principles in different conversations

Knowledge of the presence of deliberation principles in professional conversations offers opportunities for both practitioners and students in universities to learn about professional conversations and how to improve them. The results might also give rise to new research questions for qualitative investigations of a transcribed conversation, or, if present, of interviews with the participants. Moreover, if the MCD is further developed, it might be possible to aggregate the results and compare the presence of deliberative principles between medical offices, between general practitioners and doctors at hospitals, between teachers with different backgrounds and between different schools. It might also be possible to describe similarities and differences between various professional groups, a factor which in turn provides the possibility for professionals to learn from each other within a sector, across sectors and across nations.