Overcoming distrust of a method associated with business studies

6The term “focus group” which prevails in Anglophone literature [9] is often used for convenience in French. Its non-specific use is questionable because it refers to a somewhat reductionist view of the interviewing strategy. In effect, focus groups concentrate the attention of the people being interviewed on a subject in order to stimulate their responses to it. The emergence of this term as dominant was not the original intention of the man who involuntarily invented it. [10] In fact, the original technique to which the expression “focus group” refers was not so much intended to mobilise a collective dynamic as to canvass (individually or collectively) the participants in a semi-directive framework using a stimulus which was not simply a facilitating mechanism (a projective technique) but which was itself the subject of the research. In the 1940s Paul Lazarsfeld, Robert Merton and Carl Hovland were seeking to improve the study of the reactions of listeners to radio programmes broadcast during the war and how propaganda was received. In this context, the use of the term “focused” (from which we derive “focus”) designates the identification of the stimulus as the subject of the study: this is centred interviewing. Despite its specificity, this terminology then colonised the entire field of studies carried out using group interviews. However, in marketing [11] and social science surveys where stimuli have been used, most often it is not a question of directly submitting the object of the study to the interviewees but rather the elements which are associated with it, which avoids the imposition of the research issue. In this way, in contrast with centred interviewing which aims to provoke and observe attitudinal reactions (a more or less positive or negative connection by the subject to the topic studied), the purpose of collective interviews remains to understand the relationship of the interviewees to their environment by studying their discourse.

7The French expression “entretien collectif” (“group interview”, or “collective interview” in English) is therefore more satisfactory because it refers to the smallest common denominator for this data collection technique. In so doing, it makes explicit what the key difference is between this type of interview and the other generic interview category: in other words, whether the data is collected in an individual or group situation. It’s clear that in the social sciences, group interviews, just like individual interviews, can be directive, semi-directive or non-directive, may be developed in the form of life histories or even, depending on the researcher’s epistemological choices, may adopt a conversational and social approach [12] as opposed to one which probes via empathetic neutrality. [13] In other words, group interviews are first and foremost an interview, a method which consists of seeking and collecting discourse in order to understand the reasons, justifications, perceptions and thinking of the actors involved.

8We must therefore move past the reticence or indifference that has long prevailed within French social sciences because of the identification of this method with the services billed by market research and consulting groups under the name of “focus group”. [14] These services, which aim above all to spark sound-bites from participants by associating ideas or sensory responses, do not constitute interviews in the strict sense of the term but rather expressive experiences lacking a theoretical or epistemological status. Furthermore, these situations primarily consist of a succession of requests personalised for each participant by a “presenter” who systematically controls the order in which people speak, which represents an increase in bilateral situations rather than generating interaction between the participants. Once we set aside the spectre of such exercises, and finally recognise the status of interviews within the framework of methodological requirements suitable for the social sciences, it’s possible to consider what’s really at stake in the choice to seek and collect collective rather than personal discourse.

9From this perspective, it is not enough simply to claim that the group interview is a form of interview in order to justify its use. In effect, by increasing the number of factors which influence the construction of data collection in comparison to individual interviews, using collective interviews forces the researcher to make choices which have direct consequences on the situation in which the interviewees are placed, as well as on the nature of the data collected, and which influences the types of analyses that will then be possible for this data. These choices cannot therefore remain implicit. Yet it is not uncommon for publications which result from those surveys which have made use of discussion groups to include insufficient justification to explain their choice of research strategy, even though, at the same time, they include information on the socio-demographic characteristics of the members who make up the groups and the conditions governing how the interviews are carried out (place, length, period, type of recording, etc.).

The group context as research design

10It’s crucial that the mechanism for bringing together the interviewer and the interviewees not only leads to a situation in which there is a research relationship between these two types of actors, but also to the conditions of co-presence between interviewees. The analytical gains and limits which may be expected from the possibility of studying discourse produced by a collective dynamic are not independent of the conditions constituting the group which expresses itself. Justification of the criteria for the co-presence of the interviewees is required or the objective of an appropriate fit between the research question and the data collection method is compromised. It is particularly important to clarify, on the one hand, the mechanisms of the interview set-up which determine the group dynamic and, on the other hand, whether the types of interaction likely to be developed in the group allow for the production of adequate material to address the research hypotheses. When the interview method is examined in these terms, it is possible to specify the constraints within which the researcher’s choices must be justified.

11Two major conclusions can thus be set out. On the one hand, the observation unit is the group and not the individuals who make up the group. This means that even if it is possible to document features at an individual level (via filter questions during recruitment, or individual questionnaire after the interview), the fact remains that the individuals are recruited solely to form the groups and not to fulfil a particular function. On the other hand, the unit of analysis is the dynamic level of interaction (“what happens in the group”) and not the group itself in a static way. This means that the groups are formed in order to be used as context (in the sense of the configuration of the actors) allowing interactions between individuals (dynamic production).

12These two conclusions highlight the fact that, through “what happens in the discussion group”, the collective interview provides an intermediate level of analysis which consists of a (socially characterisable) context of interactions (between individuals). This meso level is situated between the micro level of the individual, which is conceived as not only a unit of observation but also as a unit of analysis in atomistic epistemology, and the macro level of social structures where, in an holistic epistemology, the individual acts as the unit of observation for the socio-demographic variables from which he or she is composed. Naturally, the definition of the notion of context is not stable and can refer to several scales. [15] It is not a question here of fixing it arbitrarily but of moving towards the specification of characteristics corresponding to the level of observation provided by the discussion group. In so doing, it becomes clear that the choices made by the researcher in forming discussion groups, far from being simple pragmatic ways of providing themselves with the means of carrying out research, are more fundamentally the criteria for defining the interaction context which determines the potential interpretive limits of data analysis. The requirements to justify the configuration of the survey (the “research design”) thus come into play.

From the number of participants to the composition of the groups: how “real” is the research context?

14In research undertaken by the business sector, the criteria for the number of participants to a group are strictly related to the practical issue of having sufficient material (in this instance, discourse) to represent a profitable unit cost. Thus, since it is difficult to envisage mobilising participants for longer than two hours, the usual recommendation is to have ten to twelve participants available in order to maximise the production of varied discourse in this time period. The impact of the number of participants on the nature of the material collected is not examined as such. Similarly, the number of groups is envisaged initially in relation to the time and financial constraints imposed by the deadlines for concluding the study, which leads to few groups being organised, often only three or four.

15Since the 1980s, Anglo-American academic literature has included a large number of books and articles specifically devoted to the ethical, epistemological and methodological aspects of group interviews. [16] In contrast to what we have observed for the focus group exercises carried out by the business sector, this literature focuses on the criteria for the number of participants but it does not really agree on an optimum strategy. Some recommendations advocate a limited number of participants in order to allow for more in-depth interaction, whilst others suggest bringing together sufficient participants so that, even if a certain number withdraw, enough active participants remain in order to maintain a collective dynamic. The standard recommendation in favour of five to ten participants is not particularly useful since it amounts to assuming that there is no difference between once or twice the number of participants within a context where the point is to maintain a collective dynamic. However, the sequences which allow members of the groups to present and adjust their reasoning mobilise in general two or three main speakers (or enunciators), whilst the other participants punctuate these sequences with a few remarks or remain on the sidelines before becoming involved in a subsequent sequence.

16The identification of this phenomenon that what people say develops through the interaction of a few individuals is important, as it forces the recognition that collective interviews using discussion groups produce a non-formalised and fluid discourse which is more like a conversation (more or less managed depending on the role assigned to the moderator). They are not, therefore, a deliberation in which the rules are gradually developed until the situation is institutionalised by the clarification of criteria for speaking, status, function and desired result. [17] The conversational aspect underlines the pertinence of the “discussion group” label and allows us to associate a conceptual value relative to the type of interaction which constitutes the collective interview: as part of a research mechanism, they remain situations created by the researcher but, through the conversational exchanges which take place, they are also situations which relate to ordinary experiences in the world around us. This factor constitutes an absolute distinction between collective interviews via discussion group and both the focus group exercises organised by the business sector on the one hand and in-depth individual interviews in social science research on the other, neither of which can ever reproduce the characteristics of the “real” world and therefore remain as strictly research relationships. These considerations about the nature of conversational exchanges within discussion groups to a greater or lesser degree lead us to the issue of the “naturalisation” or “artificialisation” of the situation. The question about the number of participants in a discussion group therefore is not only descriptive but it has consequences in terms of the realism of the data-collecting protocol.

17The notion of realism in the group situation corresponds to the academic project to account for actors’ reasons. Considering the analytical gains anticipated from this approach, the relationship between the group interview method and realism has been taken to a higher level. Certain authors have tried to theorise a research design known as the “ethnographic focus group” [18] whilst studies of the reception of the media carried out with discussion groups within cultural studies have been explained in terms of an “ethnographic turn”, [19] and there is an even clearer move towards taking a fundamentally ethnographic research approach to group interviews. [20] This desire to link the group interview method to a “realistic” field setting, as we have seen in terms of the criteria for the number of participants, is complemented in a significant way by a second criterion: how the researcher goes about forming the groups. A comparison of research designs appears to confirm that these two criteria are linked. When the group is “natural” (i.e. known to each other beforehand), there are usually three to six participants, which corresponds to the circumstances of normal life that these research designs are striving to recreate (the family group, a circle of friends, or group of colleagues, neighbours or activists). When the participants do not know each other before being brought together (and thus the realistic nature of the research set-up is diminished), there tend to be more participants in a group (always more than six).

18Indeed, whilst commercial studies guarantee the reliability of their stimulus-response protocol through the “freshness” or candour of the participants, which would undoubtedly be jeopardised by any complicity or familiarity between them and they thus never recruit participants who are already known to one another, this latter arrangement is frequently found in social science research. This contrast is an indication of the differences in the results envisaged by the professionals for each of these two sectors. Moreover, since the academic objectives of the researcher are to vary the set-up in order to avoid interpretive bias and to achieve saturation in data collection on reasoning (accumulating interviews until nothing new appears), this results in an opposing trend in social science research compared to what is observed in commercial studies: the number of groups is much higher (up to several tens of groups) and the number of participants in each group is much lower (Table 1).

Table 1

Number and size of discussion groups in research using collective interviews

Authors Theme Number of groups Participants per group Liebes & Katz, 1990 Reception of the television series Dallas in order to test the hypothesis of cultural imperialism 66 6 Billig, 1992 How the British perceive the royal family 33 [+ 30 others] 3-5 [1 (x5); 2 (x25)] Gamson, 1992 Discussion on four themes to evaluate the likelihood of individuals becoming politically engaged 37 4-6 Hunt, 1997 Understanding the “race” riots in Los Angeles in 1992 15 3-6 Press & Cole, 1999 Women’s perception of the media discourse on abortion 34 2-5 Buckingham, 2000 Perception of politics by teenagers in the United States of America and in England 18 3

Number and size of discussion groups in research using collective interviews

19If, in works in English, the objectives of diversifying the groups and achieving saturation of discourse lead researchers to accumulate a significant number of interviews, another factor undoubtedly plays a part in this. Since quantitative methods continue to dominate social psychology research, political science and, to a lesser degree, sociology, an implicit pressure is exerted in relation to statistical reliability criteria, which leads researchers to attempt to safeguard collective interviews against criticisms of representativity bias within the sample. It is sometimes even the case that concern to attain validity through intersecting variables within a random sampling approach is explicitly recognised by the researcher. Gamson’s research aimed to achieve this through recruiting a population with weak socio-economic resources but met with the problem of a refusal to take part in the survey. Since self-selection resulted in a favourable rate of response from around 10% of contacts made, the objective of representative sampling was finally abandoned in favour of a diversification strategy mindful of certain criteria such as a balance between white and non-white participants. [21] Similarly, in a survey about the perception of AIDS, the recruitment of 351 participants divided into 52 groups aimed at “covering a variety of demographic characteristics such as age, social class, ethnic and national identity” and at including various professional groups but, in the absence of data on reference populations, in fact consisted of a diversification which could have been achieved with fewer groups. [22] These diversification strategies prove to be in fact the only appropriate ones in relation to validity that research using collective interviews can aim for in terms of identifying types of reasoning, since they cannot establish how widespread it is within a reference population. [23] Moreover, researchers who do attempt hypothetically representative recruitment, do not then attempt to make generalisations concerning reasoning from their results and interpretations, but, as with other research conducted via group interview, attempt instead to characterise types of reasoning in all their complexity. This is the direction being taken by work in French, which, moreover, tends to be more at ease with working with a reduced number of groups.

What is at stake in the composition of groups

20Seen as the corollary to the criterion regarding number of participants in the context of increasing the realism of the research design, the criterion for the composition of the groups also stands on its own. Several factors play a part at this level, with anticipated consequences each time for the collective dynamic of the discourse that the members of the group may produce.

21One point that has already been mentioned concerns the nature of the links between participants: to opt either for those who already know each other (a natural group) or for people who do not know each other (an artificial group). The expected benefit of natural groups is that they do not require a special effort to encourage participants to speak and the researcher thus avoids loss of time at the start of the interview. Their familiarity with one another makes taking turns to speak more fluid and this type of group is thus more suited to non-directive or semi-directive methods which are loosely moderated. Within this framework, it is possible to rely on the relative autonomy or self-management of the discussion by the group itself, with the moderator only having to intervene in cases of obvious and prolonged digression, depending on what is provided for in the interview guide, to introduce a new theme. The material collected is in this respect similar to conversations studied within the context of ethnographic observation of the primary [24] or secondary [25] socialisation of the actors.

22In order to facilitate this process of making speaking seem natural, several organisational elements are generally adopted. In particular, recruitment is undertaken via a founding member of the group who contacts the other members so that their participation appears to them to be as a result of a social invitation. The group is also gathered together at a time and place determined by the members, often at the home of the founding member. This requires much greater availability from the researcher but it offers him or her access to aspects of the interviewees’ living conditions that would have been otherwise invisible.

23When certain material elements fail to make the research setting as natural as possible (the group is not held at the home of one of the participants, for instance), the irreducible nature of the initial request to participate can lead to the conclusion that the research relationship continues to take precedence over the normal relationships between members of the group and thus structures their interactions. Conversely, without completely reducing the bias of self-selection on the basis of a sense of competency, a highly naturalistic setting serves to improve the rate of acceptance for participating in the survey. Several cases testify to the usefulness of attempting to make the research setting as natural as possible in order to maintain the flow of the conversation and facilitate the expression of a full range of interactions. [26] Thus, for a subject which is not morally or culturally easy to tackle, “small groups of people who know each other provide an encouraging and non-confrontational setting in which the interviewees can discuss personal, controversial and potentially emotional aspects of abortion. […] [T]he group method maintains the interaction between interviewees who appear to quickly formulate their own questions as much to examine as to clarify their respective positions”. [27]

24Recruiting people who do not know each other allows for more diversification in a short time period. Beyond this, it also allows the allocation of a participant to a group according to socio-demographic characteristics. The research strategy then consists of testing hypotheses about differences in reasoning as a function of the social characteristics of the actors. This strategy brings us to a second crucial point in recruitment criteria: the homogeneity or heterogeneity of the groups. Since a totally homogenous group is inconceivable, the researcher aims for a degree of homogeneity in relation to the initial hypothesis in a deductive process which involves (in)validating a causal relationship which has been theorised prior to collecting the data. Amongst the work which revived the use of group interviews in the social sciences, the survey of how political and social information was received by different categories of the British public corresponds to this model. [28] Extracts from the television programme Nationwide were shown to 26 groups who appeared to be socio-economically and politically homogenous. The results indicate how interpretations of the same content are structurally different between the groups.

25Relatively mixed recruitment approaches are used when the composition of groups who already know each other is combined with screening, for example in relation to “cultural community” membership, [29] age or gender. Thus, in the case referred to on abortion, the groups who knew each other were either “pro-life” or “pro-choice”. The contrast between the two approaches is due more fundamentally to the research preference for either an inductive strategy, associated with the recruitment of natural groups, or for a deductive strategy which needs to control the social characteristics of the members of the group. In fact, for the groups discussing abortion, the “pro-life”/ “pro-choice” dichotomy was used to make up the groups but proved unsuitable as a variable for interpreting data because ambivalent reasoning led to a reading which was more complex than binary. [30]

26A third factor can play a part in the recruitment strategy. In some ways it is a matter of optimising the discursive dynamic by forming the groups based on similarity with the subject of the research, which tends to correspond with a composition based on individuals who know one another but can be considered for artificial groups if the selection variables allow it. Thus, for the research on the British perception of the British royal family, the survey was conducted among natural groups in order to facilitate the expression of common ideas on this subject, and, moreover, the researcher chose to use the family unit at home as being an appropriate natural group whose social identity was similar to the subject studied. This strategy aimed to stimulate discourse by relying in particular on forms of reasoning by analogy: “whenever families talk about the royal family, they talk about family; and whenever they talk about family, they talk about themselves at the same time”. [31] Applying this strategy to every subject is not easy but can be attempted. In this way, the survey about the reception of the TV series Dallas in different cultural communities around the world set up each group in the form of three pairs of friends, thus implicitly referring back to emotional life and friendships as being what drives the intrigue of the series. [32]

The contribution of interaction dynamics to explaining reasoning

28At the basis of using the group interview method is the idea that the actor does not formulate reasoning by an isolated cognitive effort but by social mechanisms which interactions within the group recreate. The more there are exchanges with others, the more the actor can work something out for himself or herself and express preferences and reasons to others. The context of the group interview (co-presence of participants who are encouraged to exchange their points of view) is thus suitable for collecting opinions and reasoning. The hypothesis can even be formulated that, in the course of these interactions, specific or sensitive moments in particular will reveal how opinions are formed. [35]

29To analyse interactions, it is useful to draw up a typology of discursive interaction dynamics. From observing the groups in the two approaches which have been presented, seven dynamics have been identified: adhesion, confirmation, solicitation, questioning, speculation, distrust, contradiction. Adhesion consists of signalling agreement with a view held by another participant. Confirmation deepens adhesion in that it consists of arguing agreement, often by taking the original idea further. In contrast, distrust consists of signalling disagreement and contradiction consists of arguing disagreement. Furthermore, three co-operative dynamics may complement one another. Solicitation consists of asking others what they think about a point of view. Questioning consists of asking someone to explain their point of view. Finally, speculation consists of commenting on a point of view in order to evaluate its impact.

30These seven types can be arranged in three categories depending on whether they are dynamics of convergence (adhesion, confirmation), divergence (distrust, contradiction) or neutral co-operation (solicitation, questioning, speculation).

Table 2

Categorisation of discursive dynamics

Convergence Neutral co-operation Divergence Adhesion Confirmation Solicitation Questioning Speculation Distrust Contradiction

Categorisation of discursive dynamics

31It is through these dealings with the other members of the group that each participant is led to confront, amend or revise their thinking. Without seeking to question their significance, particularly in terms of understanding the mechanisms of interaction within the groups, the three dynamics of neutral co-operation will not be examined within the scope of this article, for two reasons: one practical – analysing them requires more space than is available for the remainder of this article; one substantial, in that the available literature focuses particularly on the distinction between the mechanisms covered by the two other categories of our typology and it will be more instructive to situate the results derived from our analyses in relation to these considerations.

32Divergent dynamics provide an interaction framework generally considered as very favourable for explaining the reasoning of the participants. By pushing speakers to defend their point of view, they produce a spur for clarification and specification. This is why pinpointing and taking on board divergences analytically is supported by discourse analysis theories [36] and can form the basis of a research strategy for group interviews. [37] It remains to be seen to what point divergences are taken on board because it is likely that a low intensity of divergence will not be associated with more in-depth interaction. A sequence with a group of young men in 2007 makes it possible to see the difficulties in developing the interaction when the intensity of the divergence is low, as the use of (bold) indicates below, and without significant argumentative impact (isolated connectors).

33

Philippe – When you really look, Ségolène Royal is also up with the fairies, eh… Because, er… the creation of juvenile units in the police [brigades des mineurs] er… they exaggerate… because below my house, there’s one, so she needs to catch up a bit… She wants to reform: ah, I’m going to create a juvenile unit. Er, it already exists!
Greg – Yeh but, on the other hand, her programme for reinstating community policing isn’t bad, all that…
Charles – Yeh, for me, community policing, I don’t think that it’s too…
Greg – Well, to give the cops the means to…
Charles – But what means… Yeh, they’re going to answer on the first call…
Greg – Well, Subarus, tasers, flash balls [laughs]… It’s like Christmas, eh…
Damien – They’re already over equipped…
Greg – No but, community policing, they know the blokes and that’s it… It’s easier like that…
Charles – And for minors on the other hand, I agree with Sarko about repeat offenders. Because young people have changed…

34The differences in preferences around the issue of security come to the surface but their reasoning is not explained and the sequence results in a juxtaposition of elements which do not structure an argued discourse. According to the categories of our typology, the dynamic of distrust is not followed by a dynamic of contradiction. The participants do not feel any embarrassment about expressing themselves, so it is not potential inhibition about a topic that is considered as sensitive which explains the absence of in-depth discussion but rather the absence of intensity in the confrontation of their ideas. In comparison, in a survey about their relationship to politics, the residents of a home for young workers accept a deeper divergence where each of the participants involved develops their point of view and consolidates it as they confront disagreement. The divergence markers (in bold) include pronouns designating the speakers (I/You) which demonstrates the engagement in the interaction. This intensification of divergence is reinforced by argumentative attempts which ensure that the dynamic of distrust moves on to a dynamic of contradiction and results in the clarification and specification of opinions.

35

Marion – And if you are not up to date, how can you expect to be… how can you be for or against?
Hocine – You know that ignorance of the law is no excuse.
Marion – But you know that if, each time…
Hocine – Each time that a law is brought in, it goes into the Official Journal.
Hamed – Yeh, well I don’t read it every day [laughs].
Hocine – No, me neither… The law on hunting, I’m not a hunter, eh… No, what I mean is that, er… people who make the laws of a country are people who are supposed to have been elected by the people.
Marion – Yes but that doesn’t mean that everyone agrees with them about everything. Hey, you’re not going to tell me that these people have been to see each citizen in the street to ask them. You’re not going to tell me that. And people can think for themselves. And they haven’t been to see each citizen to ask them “do you agree with this law?”
Hocine – No OK but…
Marion – And it’s not logical when it concerns us directly when, for instance, the guy from the Parliament who’s going to decide, he doesn’t care… OK, he’s a citizen too. But it isn’t because he’s step above us that we shouldn’t have a referendum, you see! So I don’t think it’s logical that we don’t have more referendums…
Nadia – You want to have referendums on everything?
Marion – No, not all the time but…
Hocine – He’s not… He’s not… He is not a citizen above you. A member of Parliament has as many rights as you…
Marion – No he doesn’t! He makes laws without asking our agreement so he is above us.
Hocine – But you’re supposed to have elected him.
Marion – Yes, we elected them but, afterwards, we’ve got nothing left to say?!
Nadia – It’s… it’s true that it’s difficult to see how all these laws get passed. To see that we have an impact on these laws which are passed, er… in the National Assembly and which are applied afterwards… in the street, by the police or… You see, you don’t feel as though have any control over it, when you are an average citizen.
Hocine – No but, it’s true… but even the average member of Parliament, he… he realises that he doesn’t have a significant impact. I knew a member of Parliament who…er, who…the former mayor of my village… And even he didn’t feel that he carried any weight in the votes in Parliament.
Nadia – You see, so that means that you’re in a set-up where you don’t know what’s going on… Marion – And then, you know, with a representative like that… eh, we are not represented very well!

36Comparing these two sequences is not sufficient to demonstrate the general trend according to which the research design in which participants know each other beforehand facilitates the development and deepening of divergent interaction. However the fact is that, without a specific prompt to guide the participants to take on board their disagreements, the research design for interviews around the 2007 campaign did not reveal this type of dynamic between participants who nonetheless had different preferences. This ties in with the findings of an experiment on the reluctance to expose opinions to strangers within the context of discussion groups. [38] As the majority of research which mobilises the “natural groups” previously discussed proves, knowing each other beforehand can enable participants to go beyond inhibitions regarding how others will judge them: since any eventual disagreements are not powerful enough to threaten the links between them, they can speak without fear of prejudice. In groups of political activists in the survey into people’s relationship to politics, moreover, familiarity with controversy leads to coded expressions of mutual respect after an exchange of opposing views.

37Convergent dynamics are considered intuitively as less useful for analysing the formation of opinions and reasoning than divergent dynamics. One element that is often mobilised to explain this but is never really demonstrated consists of assuming that convergent viewpoints lead to things remaining unspoken which means that participants do not feel the need to employ explanatory tactics. A closer examination of the groups who already knew each other in the survey on relationships to politics demonstrates, however, that convergent dynamics can be useful in explaining and deepening reasoning. For example, in the group composed of Black friends in employment, the repeated appearance (indicated in bold) of convergent opinions about their place in society even led to a statement setting out causal theories concerning politics.

38

Jean – But you must… You must, I think, you must still keep up to date, especially with the life that we’re living now. Because… for fear of… well, we risk… er, because our politicians… these people, they aren’t only there to lead us but to order us around…
Achille – Oh, that… that’s… that’s been going on for years…
Jean – because we don’t tend to pay attention. Unfortunately, we have to really pay attention.
Gérald – He’s absolutely right, eh… Because, in fact, if we look at the society that we are in… er well, the rich are getting richer and richer and the poor grow poorer. And this is a political strategy, because, actually, when we look at the elite group of politicians, we know very well that it is always the same. And it’s their children who are going to take over the reins.
Achille – It has always been like that… They run everything… And it’s been like that since… pfff… forever.
Jean – Exactly.
Gérald – Well, when you look, when you look, ultimately, the laws that they make, it’s not for them! It’s for us, the average people and the poor people, the ordinary people in society. So, what’s happening? For example, take a person who has made a mistake, fine, OK, but he gets a penalty notice, a large fine and he doesn’t have the money to pay it. And the Home Secretary, what does he do? He starts by saying that, that… that the law enforcement agencies must be stricter, that is to say that they work together as much as possible. But just watch… he doesn’t get a penalty notice, the guy who makes the law, just because of that, even if he commits an offence like we see, like we’ve seen… And even for those who get a penalty notice, even if there is an offence, the rich can pay, they can get out of it… But the poor, what are they going to do?… They can give everything that they have and they can’t put any money away, they can’t invest or buy. They’re always going to be poor… It’s deliberate, ultimately, it’s deliberate… And it’s not only the penalty notices but everything… They put pressure on everything. Everything is worked out in advance. We… the… the, how shall I put it, the poor people…
Jean – [smiling] The proletarians… the proletarians of the Republic…
Gérald – [laughing] Yeh, the proletarians of the Republic… us… us… what can we do?
Achille – Well, we understood the system but we can’t do much about it.

39At the end of the sequence, the spontaneous creation of the expression “proletarians of the Republic” helps the person speaking who is searching (quite literally) “how to put” his opinion. By associating the terms which exactly combine the thinking about the link between political domination and socio-economic status – a thought process which itself prolongs the suggestion that started the sequence – this intervention indicates that the dynamic of confirmation has been used to co-construct meaning. The dynamic of convergence when pushed to its limit thus provides the researcher with a discourse which clarifies the opinions of the actors. The same process is observed for students at business school and indicates the terms of their distrust of those who govern (a demand for proximity and authenticity), as well as for secondary school teachers when they develop the analogy between their dissatisfaction towards their educational establishment and towards political institutions (a demand that the decision-makers take into account the point of view of the actors on the ground). Across the board, and with even higher intensity in groups of political activists than for politically unengaged citizens (a counter-intuitive result in relation to the starting hypothesis), what comes out of the dynamics of convergence on the subject of political representation is the expression of high expectations and high requirements.

40The hope of revitalising the representative link rests on the articulation of two generic types of reasoning: on the one hand, the division of labour between the representatives and the represented who appointed them is seen as a satisfactory principle of political organisation and, on the other hand, the failings of representative democracy are attributed to the desire of politicians to withdraw from the representative link. The demand to rebuild the representative link also rests on the articulation of two generic types of reasoning: on the one hand, it expects the representative to conform to procedural requirements (respect campaign commitments, engage with their remit, clarify divergences) and, on the other hand, they are expected to demonstrate the representative link on a regular basis by their presence, availability and willingness to listen, without this exonerating them from making public policy. This explains the strong symbolic investment in and attachment to the cultural conditions of political legitimacy, which at the same time appears to be implicitly allied to a sense of mourning for the effectiveness of public policy, or, with a certain fatalism, a tolerance with regards to its lack of effectiveness. [39] Without being generalisable, these results stemming from an inductive approach reveal a level of complexity and development of opinions which require us to examine conceptions of political representation among citizens more closely. [40]

What can be expected from groups formed by controlling variables

41The groups formed to consider the 2007 presidential election, recruited according to sociodemographic characteristics without the participants knowing each other beforehand, did not deploy strategies of explaining and specifying their reasoning as extensively as the survey groups around relationships to politics did. Convergent or divergent elements tended to be limited to brief indications of adhesion or distrust and were without discursive consequences which would have enabled a move towards more in-depth dynamics of confirmation or contradiction (Table 3).

Table 3

Mobilisation of interaction dynamics according to survey configuration

Type of group Convergence Impartial collaboration Divergence adhesion confirmation solicitation questioning speculation distrust contradiction natural ++ + + + – ++ + artificial + – – + + + –

Mobilisation of interaction dynamics according to survey configuration

42The contrast presented by this summary table of group dynamics according to their natural or artificial composition does not mean that the groups where the participants do not know each other are unable to provide a result. However, it is a question of treating the discourse in a manner appropriate to the research design.

43Obviously, recruiting those who already know each other may produce homogenous groups in a certain respect, which in turn produces a variable that is useful for interpreting the data, without this having been intended. Thus the experience of discrimination due to skin colour influences the thinking of the group of Black employed friends. The extract cited above focuses on their perception of distance between themselves and the political leaders and is clearly confirmed by other sequences in their group’s discussion which particularly mobilises the idea of minorities in relation to how society functions in general; specifically political representation and more specifically still the use of referenda, perceived as a process wherein a circumstantial majority is likely to crush different minorities. The significance of reasoning in terms of variables is thus not to be dismissed even in the case of groups who already know each other, but it is less reliable when there is a lack of systematic control between the groups.

44When research is designed to test reasoning in terms of variables, it is preferable to control for these variables through the composition of the groups. The technique for group interviews around the campaign in 2007, for example, aimed at testing a certain number of hypotheses within a deductive approach. For instance, the desire to test opinions about the possible election of a woman as president led to forming groups of young people based on gender. In the event the young men neither commented on the hypothesis of the election of a woman nor expressed any opinion about the candidate as a woman, whereas the group of young women did. Moreover, this allowed us to trace the development of opinions from a concern by two members of the group to justify casting their vote (amongst other reasons) on the basis of a need to affirm the ability of women to take on political responsibilities, via irritation with the personality of the candidate, who was in the end perceived as being harmful to the credibility of women, and finally to the denunciation of what was perceived as her instrumentalisation of female identity. The collective expression of increasingly unfavourable opinions of the candidate created moreover a context of emulation which enabled those who had initially supported this candidate as a woman to voice their disappointment.

45In the survey about the 2007 presidential election, the main hypothesis focused on the differences in engagement with the campaign according to the socio-economic status of the individuals. The results confirm the differences but neither in the ways expected nor for the reasons envisaged. Although the link between socio-professional status and the level of political competence was confirmed, it did not result in a better ability to make sense of the campaign in the group of advantaged socio-professional categories (SPC+); in any event in so far as this ability was tested in relation to televised news (around environment, employment and security issues) during the second meeting with the groups. Even though the 2007 campaign was characterised by a rapid rotation of the most salient themes, it is striking that with all the groups there was a lack of sensitivity to the issues presented to them, whilst their reasoning was closely linked to the candidates’ personalities. Thus, even after introducing the theme of the environment a second time, it was not developed further by any of the groups, and the majority of the comments concentrated on the personnality of Nicolas Hulot (following a report about an event in support of his Pacte écologique). Despite this, it was possible to identify how socio-economic status affected the relationship to issues, which the group interview method, as a result of the homogeneous composition of the groups, helped to bring to light.

46Of the discussions studied, overall it is clear that public policy issues could not on their own make sense of participants’ relationship to the campaign programme, which turned out to be dominated by an overwhelming focus on the candidates’ personalities. One could hypothesise that, just as in theories of political competence or sophistication, the level of knowledge and interest in politics, which is higher in the SPC+ category than in the disadvantaged socio-professional categories (SPC-) and for young people, provides the members of the former group with alternative resources for following the campaign so that they can “intelligently” construct a perspective on the election issues or the candidates’ manifestos. Conversely, it is amongst the youth and the SPC- category that we see, in an incidental and irregular way, the development of a certain amount of political reasoning relating to congruency between the election issues and the life experiences of the members of the groups. This is particularly the case for concerns about their material conditions of life, [41] whereas this compensating strategy of engagement with the election issues does not occur amongst the SPC+ category. Thus, amongst the young women, a police woman initiated a sequence about security and another, an unemployed women, a sequence about the incongruity of making overtime exempt from tax when many people do not have a job. Amongst the SPC-category, an employee at a hairdressing salon initiated a sequence about the proposal for a minimum wage of €1500 by referring to his level of salary. Similarly, whilst the topic of employment measures was hardly discussed in this group, as it was in the other groups in relation to the candidates’ reactions to Airbus’s redundancy programme (which suddenly became a salient issue in the campaign and which was shown to the groups in a report on the televised news), it was eventually tackled in relation to personal situations. The following extract demonstrates this on the subject of Nicolas Sarkozy’s proposal to stop benefits for job seekers who refused two successive job offers.

47

Béatrice – From what they said, er…the right has picked up social programmes, so now, right/left, er…
Benoît – Social, well, OK… on the right there is more inflexibility all the same. For example, with Sarko, if you refuse two jobs when you’re unemployed, he strikes you off, eh… So, if you call that resuming social programmes…
Corinne – Well it also all depends on what they offer you because… […] I remember… at the moment, I’m looking for a job. Over the years, I’d done lots of temp work and then I managed to work. So after that, I put on a different hat: I was employed on a fixed long-term contract. So, my contract was coming to an end and I had to get back into the fray and, at the time, I remembered, it was when they had created the infamous PARE plan [back-to-work assistance plan]. After everything that I had heard, at least understood, I thought that they would offer me a job. I was, what, 25 years old, I’d finished my studies and I’d worked so I was entitled to unemployment benefit. I went to the job centre all happy: so, what have you got to offer me by way of a job? They had nothing to offer me. I really would have liked… Now, Sarko, he’s right, there must be people who take advantage of the system. I did, others did too. There are times, if you don’t have any unemployment benefit, it’s true that maybe you’d make more of an effort. I’d like it if they helped me to find a job, if they said, “there you go”… But it’s got to be right. Nowadays, I can’t be too demanding about the salary or the job. But if they offered me something that really didn’t have anything to do with what I do, what I’m looking for, it’s true that that’s still important… From my point of view, if Sarko gets in it’s a bit worrying all the same.
Béatrice – But when you see certain people who get income support as a result of being unemployed and are moonlighting. Well, ultimately, whether it’s for income support or unemployment benefit, if they told them that, there would be less people moonlighting.
Corinne – Yeh, that’s true but you must also see…
Véronique – Do you know people who do that?
Corinne – You see, at the job centre, they have to help people…
Béatrice – I know people who are on income support, who have a 3-room place on housing benefit, who don’t have any kids to look after and, at the end of the day, well they earn more than me. Yes, they earn more than me. I don’t even earn the same amount of salary as they make in handouts.
Corinne – But there are lots of people like that, who don’t pay any taxes, or rent etc. I’ve got a friend who is younger than me, who works in a clothes shop in Châtelet, well he does that. But I said to him: think about your retirement. But, despite everything, the job centre, I’m sorry, but when I went there, I really wanted to find a job, you go there…and there wasn’t just me. Really, you know, sometimes we criticise the immigrants but they’re not there to loaf around.
Véronique – No, no, that’s for sure…
Corinne – They’re there to look for a job so the critics should be quiet. And you go there and you leave, you don’t have an appointment. We say to them: give us an appointment. They offer us an appointment in three months! Shit…three months! And they’re there with no worries… Ah, they’re not panicking, eh… Ah, the pace is… No, but it’s not that: the job centre, it’s useless. No, it’s useless. On the other hand, if Sarko gets in, maybe he’ll make them do something. That’s really…
Benoît – And what’s he going to do with the people he’s going to strike off after two offers. Well? He’s going to leave them with nothing. What’s that going to give all those people?
Béatrice – Well, the street…
Benoît – We don’t talk about that, do we?… It’s true, we don’t talk about it. What’s going to happen to those people who will be skint, who were living on their unemployment benefit?

48In light of what has just been demonstrated, the interest of the group interview consists in providing the researcher with data which is not produced from hypotheses alone. This particularly enriches deductive strategies which, through the criteria for the composition of discussion groups, provide the means of testing the effects of the variables selected whilst leaving room for what the interviewees themselves produce during the course of their interactions. In this case, analysis of the discussion dynamics shows, on the one hand, a high degree of distance from the campaign issues in all groups and, on the other hand, the ability to take on board the issues, not because of formal competence but because of an involvement which comes from an articulation between individuals’ circumstances and propositions made by the candidates. This mechanism of involvement had not been hypothesised at the outset, but could be established from its recurring appearance and unequal distribution between the groups.

49* * *

50At the end of this examination of the possibilities opened up by the group interview method for analysing people’s political reasoning, two conclusions can be drawn. Firstly, from a methodological point of view and in line with the findings from the international literature, it is clearly necessary to be rigorous in planning the research design. In particular, choices concerning the composition of the groups have significant consequences for the discursive dynamics which affect the fit between the research question and the type of data collected. In this respect, we have confirmed that the most valid unit of analysis in exploiting the data from groups is the level of interaction between the participants, most often linked with the notion of thematic or argumentative sequence.

51Secondly, since both are largely a question of the relationship with politics, what comes out of the two studies which have been used here is that the discourse produced by groups constitutes complex material and cannot be reduced to a single dimension (political representation is both desired and criticised; the election campaign does not make sense either by or for itself). Even in the case of a deductive survey strategy based on a protocol which controls variables, the results are found to be out of step with the formulation of the initial hypotheses and required a theoretical adjustment in order to interpret them. In the case of deductive survey strategy based on a protocol which arranges settings which are as naturalistic as possible for participants to produce their discourse, it is even clearer that the treatment and interpretation of the data requires great caution because it provides ways into several interlinked dimensions (social stratification, cultural identity, normative preferences, interview situation).

52In fact, the discursive dynamics by which actors are led to make continual adjustments to the development of their opinions and reasoning provide the researcher with material that is unparalleled in other data collection methods. The discussion groups give access to opinions as social constructions. However this “thick” layer of discourse is only an asset if the analytical framework which is applied to it develops from the justifications which are deemed necessary by the complexity of the material. Without this, the depth becomes a mass, and although it is still possible to extract aesthetically striking illustrations (rotation of the sequence of speaking, familiarity of language, intensification of exchanges etc.) these elements have lost their status as academic data. This problem takes us to the issue of how to mobilise and present the data in publication formats which are not well suited to the inclusion of long quotations from an interview, when it is only through using the data generated by discursive dynamics (which requires space) and making comparisons (which requires more space) that evidence can be presented.