Natural Interactions in Artificial Situations : Focus Groups as an Active Social Experiment

The focus group interviews can provide a unique access to interaction ‘at play’, and can as such serve as a method for investigating the social processes in society (Holstein & Gubrium, 1995, Morgan & Spanish 1984, Meyers & Macnaghten, 1999). Power is an imminent part of the dialogue in interviews (Atkinson & Silverman, 1997; Gubrium & KoroLjungberg, 2005; Kvale, 2006), and in focus groups power relations are even more interwoven with participant interactions than in individual interviews, and are no longer confined to the relation between interviewer and interviewed (Hofmeyer & Scott, 2007). As such focus group data could be framed as on one side a biased by the social setting of the group. On the other side the participant interaction could be viewed as unique data that lets the researcher follow the construction of data. In other words, the focus group interaction can be observed as it were “talk” within a ethnographic study. This dissolves the distinction between interview data and observational data (Halkier, 2010), and as I will argue, between interview-data, observational data and experimental data. More positivistic as well as constructivist approaches do however struggle with what kind of validity focus group data has. This is reflected in questions as to what degree focus group discussions reflect real life situations, and, what can be considered a finding and what is a bias.


Introduction
The focus group interviews can provide a unique access to interaction 'at play', and can as such serve as a method for investigating the social processes in society (Holstein & Gubrium, 1995, Morgan & Spanish 1984, Meyers & Macnaghten, 1999).Power is an imminent part of the dialogue in interviews (Atkinson & Silverman, 1997;Gubrium & Koro-Ljungberg, 2005;Kvale, 2006), and in focus groups power relations are even more interwoven with participant interactions than in individual interviews, and are no longer confined to the relation between interviewer and interviewed (Hofmeyer & Scott, 2007).As such focus group data could be framed as on one side a biased by the social setting of the group.On the other side the participant interaction could be viewed as unique data that lets the researcher follow the construction of data.In other words, the focus group interaction can be observed as it were "talk" within a ethnographic study.This dissolves the distinction between interview data and observational data (Halkier, 2010), and as I will argue, between interview-data, observational data and experimental data.More positivistic as well as constructivist approaches do however struggle with what kind of validity focus group data has.This is reflected in questions as to what degree focus group discussions reflect real life situations, and, what can be considered a finding and what is a bias.This chapter discuss the question of how the validity of focus group data can be reframed when approaching focus groups as social experiments in a science and technology approach.By using this frame we first of all comes to perceive the focus group discussion as an artificial situation, while the interactions going on in the group can be described as natural occurring data (cf.Silverman, 2007).Thus this approach comes to terms with some of the problems addressed within both positivistic as well as constructivist uses of focus group methods.Secondly, framing focus groups as social experiments also highlights possibilities of a more active use of groups (by intervention) that resembles the interviewing situations as an active ethnomethodological breaching.It is within this framework of "stimulated or irritated" natural occurring data that focus groups will be discussed.
The use of metaphors like experiments, laboratory-like settings and artificial interactions is not new to the focus group literature (Morrison, 1998;Dimitriadis & Weis, 2001;Lezaun, 2007).When Merton and Kendall in 1946 presented the focus group interview as a sociological method it was with the intention of dealing with the uncertainties of intervening effects in social experiment and thus arguing that focus groups could be a relatively controlled setting for testing out norms and values in humans (Merton and Kendall, 1946).The experimental setting of the focus group was, as Merton recalls in a late paper, used in media studies, to produce observable outcomes as much as it was an interview technique (Merton, 1987).More recently Dimitriadis and Weis' raised a central point about how the focus group should be seen as a laboratory where we as moderators actively influence the group and slip into a mixed role as teacher or mentor.Lezaun argues that the moderator's skilled facilitating of individual standpoints combined with the clinical interview rooms with one-way-mirrors becomes a "machinery" that produces individual opinions.He argues critically that the focus group, primarily in market research, makes use of this unnatural element in order to produce sincere and non-socially biased opinions.This article argues, in opposition to Lezauns argument, that the researcher must not only put emphasis on the relations between people, actions, statements and occurrences in the group, but should creatively use group composition as well as topics and moderation (i.e. the control of the dialogue) to create interesting focus group data.The framing of the focus group study as a social experiment can be an epistemologically productive position, if the researcher acknowledges a), the plastic laboratory-like focus group setting, the social experiment can become a more active part of the methodology, and b) in the same epistemological position social interaction as occurring naturally within the setting of the experiment.It is, thus, not argued that the researcher should try to control the experiment in as many possible aspects (as is the notion in e.g.randomized controlled trials), but more be acknowledged as a way to make the focus group participants interested in the topic.The method becomes less a "testing tool" for values and moral, in the way that Merton created the focused interview, and more a tool for describing other versions of a given phenomenon.This argument is based on the assumption that if the focus group data cannot be argued to mimic a naturally occurring situation, then research can use focus groups for investigating aspects of the phenomenon that will become less visible outside the scientific setting of the focus group.The focus group is, in line with this argument, not directly a method that can solve a problem of knowledge by producing specific data (Despret, 2004;Law, 2004;Stengers, 2000).Instead, focus groups can be able to change the problem by bringing new versions of the phenomenon into light.Thus, it can be an active tool for producing more contrasts.In addition, the metaphor of the experiment draws attention to the ethical considerations involved in using the social setting for data production more actively.This article discusses the above themes by using data from a qualitative study on youth and alcohol based on thirty-seven focus group interviews. 1 The teenagers were first interviewed in the 8 th grade (aged 14), and then re-interviewed in the 9 th and 10 th grades.In this way, each individual participated in up to five focus groups over the three years, but within different settings.The social dynamics unfolding in studies with groups of teenagers is often argued to be very strong (e.g. Green & Hart, 1998;Allen, 2005;Fingerson, 2005) and could as such be argued to by leading to more socially biased or constructed data.The large number of different focus groups on the same topic enables this paper to investigate an active strategy of using different "social experiments" in knowledge production.

From representations to observations of presentations
What does the focus group say about actual practice?The nature of interactional processes in the focus groups can in a social constructivist or in an interactional perspective be observed as representations of reality.The focus group is a unique method in that it can account for the ongoing processes, which -through interaction -unfold before the moderator as the interview progresses.The processes, interactions, discussions and powerrelations established between the participants can become basis for the researcher's method (Tanggard, 2007).Thus, the pivotal point of using the focus group is to recognize that the statements of an individual participant in the focus group are as much a product of the processes going on in the group as they are an expression of individual experiences.The individuals' presentations (Goffman, 1959) in the group can as such be viewed as socially negotiated representations of reality.If we turn towards the suggested understanding of the focus group as a social experiment, however, we could to a lesser degree perceive epistemologically the interaction in the group as re-presentations of a negotiated outside reality, but instead as presentations in the setting of the focus group.As such the focus group can be argued to produce data of the present; we can ethnographically follow what goes on in the interview.The data observed in the group as such changes character from somehow unclear representations to become presentations when the analytical setting is moved from the outside of the group to the inside of what happens in the group.The interaction practices are not represented in the interview but actually taking place during the interview.However, the focus group setting does not exist naturally.This brings us to the central question of this paper.How can we account for the awkward combination of natural occurring data (the interactions taking place here and now) and the artificial situation (the focus group as a social experiment)?
The traditional psychological experiment had the assumption that it is possible to test human or animal behavior by narrowing down the options of possible behavior (Barkan, 1996).These assumptions are abandoned by the modern randomized experiment, where control groups and statistics account for the finding of causal effects (Kristiansen & Mooney, 2004).In the birth of the sociological tradition, experiments were regarded as making reliable data on society (e.g.positivistic concepts of social laws) (Brearley, 1931).But since the rise of Verstehende Soziologie the significance of the experiment was reduced in sociology (Gross & Krohn, 2005).Social laws were generally abandoned as a central epistemological approach (Dehue, 2005).It was also questioned if it was possible to actually test social phenomena because of their complexities and the lack of controlled, experimental settings (Angell, 1932;Merton & Kendall, 1946).In addition, the ethical questions of experimenting with people's lives were also a central issue.
In Science and Technology Studies the experiment has been closely related to the laboratory as the concrete setting of knowledge production.Karin Knorr Cetina argues that "Laboratories are based upon the premise that objects are not fixed entities that have to be taken "as they are" or left by themselves" (Cetina, 1999).At first the researcher does not have to "accept" the object as it is, but can transform it (as particles in fog chambers are transformed into traces), and second, the researcher does not need to work with the object where it is (in its natural setting a particle will in some cases be to complex, it can e.g.be isolated) and last, the researcher does not need to study the event when it happens (e.g. when a CERN particle accelerator simulates the moment of the big-bang).These descriptions fits overall on the focus group: it is a laboratory-like setting that constitutes the framing that transforms naturally existing data into a social experiment.This is a transformation of the interaction by the use of a different setting related to time and space.The focus group is as such a range of different techniques that transform natural interactions into a form, that assembles the social experiment.This makes the focus group a specific "inscription device" (Latour and Woolgar, 1979); a device that transforms first natural interactions into a form that enables the researcher to observe interactions in a way that makes us able to stabilize results in time and space (Barry 1995).The focus groups "laboratory" makes observations of social experiments possible through its technology2 of video-taping, one-way mirrors, discussion guides, conference tables and transcripts.The laboratory process will be a transcription device that makes certain natural occurring data observable as statements and transcripts.

The passive experiment
Even though most focus group studies use the rhetoric from the controlled experimental, the technologies are seldom regarded as partaking in the transformation of data.The methodological concepts of control groups and test groups are used in order to narrow down the risk of possible misinterpretation based on social interaction during the focus groups. 3In the same way, the use of stimuli material in focus groups is referring to an understanding of a stimuli-response relation in a closed setting.These technologies in the focus group methodology does as such support an epistemological assumption of the controlled social experiment.
The use of the controlled experimental setting for focus groups can be understood by what Andrew Barry (1995) has termed a passive technique.By accepting the focus group to be passive the researcher chooses not to understand it as acting upon the phenomenon that it describes.In other words, there is a tendency among researchers, when using the most controlled form of data collection, to describe the method as the most passive.If we use the passive understanding of the focus group we in the same time perceive how data is influenced by the setting of the group to be a bias.This means that such an approach supposes that focus group data is very close to natural occurring data, that may be "disturbed" by certain biases.There is a risk that the metaphor of the "natural" presents the researcher and focus group moderator as what Haraway (1998) has ironically termed a "modest witness": a scientist who through his observations only modestly describes the world.The focus group study is anything but a modest observation of facts.Indeed, Burman (2001) argues that a focus group to the participants always feels as "entering unfamiliar territory".We as researchers do often need to familiarize participants with the concept of the focus group (by introduction) and with how to engage in discussions (by moderation) (Morrison, 1998;Lezaun, 2007).It becomes a non-natural formal social space with certain expectations on group participants.This means that there is a tendency to produce rather restricted data.In particular, it is a social setting where the often non-spoken rules exclude some kind of data (c.f.Despret, 2004).
I suggest acknowledging the artificiality of the focus group situation and as such view its techniques as an inscription device that co-produces data.This framing of focus groups methods puts emphasis on the virtue of concrete descriptions of methods and process, because it allows other researchers follow how it produces knowledge.At the same time the reframing of the focus groups as social experiments allows a more actively open strategy that makes researchers produce interesting data.I will now describe this by using the concept "active experiments".

The active experiment
Harold Garfinkel (1967, p.37) asks "what can be done to make trouble?":The ethnomethodologist must, by way of a friction or intervention, make elements of a phenomenon visible which are normally hidden in natural social settings.Garfinkel terms this a process of breaching: a way of gaining insights into aspects of lives which the qualitative interview probably would not give access to.Ethnomethodology finds ways to intervene in daily life (Garfinkel, 2002) in order to produce these breaches.However, the key strategy of the focus group is not to intervene into the common practices, but to produce an artificial or experimental-like situation that can establish productive interactions with the studied (Callon, 2006).I will discuss a case well known in Science and Technology studies to open up the discussion on how focus groups as experiments can become active in producing data on the entities studied.
In a discussion of a study of sheep (sic), Vinciane Despret (2005) argues along Certina (1999), Isabelle Stengers (2000) and Bruno Latour (2000) that the construction of the experiment is decisive for the kind of empirical data that is produced.Earlier studies of sheep behavior had focused on how sheep's behavior in the flock was affected by shortage of food.In Despret's experiment, however, an extra bowl of food was given to the animals so there were more bowls of food than animals.This implied that shortages of food were no longer a dominant factor in the observations.Most studies on sheep behavior have concluded that the herd is organized according to strength because strength is central in order to get food.In supplying plenty of food, the experiment attempted to shed light on other aspects of the behavior of the sheep.The sheep no longer only behaved strategically in their fight for food, but also organized their relations according to other principles.I will not go into details with the findings; I just want to point out that this study shows that sheep have an emotional relationship and that they do care for each other by making pairs (couplings), groups etc.
Even though there is a long way from experiments with sheep to focus groups, and I have absolutely no intention of suggesting that group participants are comparable with sheep, there is a methodological finding that makes the sheep-study relevant to focus groups.By strategically making the focus group differ from natural situations, it is possible to make other parts of the researched phenomenon visual to the researcher.Thus, focus groups can be used as a kind of ethnomethodological breaching or experiment, drawing out aspects which would otherwise have remained hidden.Whereas ethnomethodology is a way of making a phenomenon visible in the natural setting of everyday life, conducting focus groups is rather a matter of taking the phenomenon out of its embeddedness in everyday life.What I suggest is to try to actively establish interactions with the studied that enable it to speak up (Callon, 2006Latour, 2004;Stengers, 2000).The social experiment can be usedas already argued -as a form of enacting of reality and thus to pinpoint that it is not naturally occurring.This enacting is essential in order to actively use the method to bring aspects of the studied forward (at show and on the stage) and in that way make it interesting -both to those that are being researched (participants in the focus group) as well as to the research society.It is not only essential that the phenomenon become interesting to the researcher -as when she is able to visualize logics of the phenomenon -but it is equally important to make the studied (focus group participant) participate with interest in the subject studied.As such the argument of using the experiments of the focus groups actively is used in a slightly different way compared to how it is presented in "The Active Interview" (Holstein and Gubrium, 1995). 4  In line with the discussion of experiments with sheep it is essential that the scientific setting is active in making versions of the studied, that allow the hitherto mute to speak up.By creating a situation which differs from the everyday life of the participants, we can provoke statements and reactions about topics such as drinking alcohol, that would probably remain hidden to the subject in an everyday life context.The group interactions create the opportunity to investigate other versions of the same reality.The parallel to the social experiment should be regarded as central in constructing social data even though it does not mean that the researcher is able to control the setting.However, focus groups produce an intervention or shock to the social interactions by pushing group participants into orally expressing their thoughts on the subject, their positions in relation to others, their view on other participants' habits, etc.In light of this, I argue in line with Andrew Barry that "What is true is that which can be seen or can be made visible" (Barry, 1995, p. 54).When pushing the social interactions out of their ordinary settings we are able to investigate some of the other forms they might take.These forms of the phenomena are what Despret terms versions, which become very central contrasts in order to analyze yet again other versions of the phenomena.It is exactly because of the non-natural setting of 4 The active interview technique successfully developed by Holstein and Gubrium is very central to sociology in its refined approach to understanding the interactions and power relations between interviewer and participant.As such they argue that interviews should be regarded as a social construction and less as a data collection process where the professional clean questions and related crisp answers could describe practice.The perspectives developed in the constructivist Active Interview approach are indeed relevant in the suggest approach to focus groups as social experiments suggested here.And they do by all means permeate the discussion in this paper.However, I suggest that the term active also could be used as a way to highlight how we can use the experimental focus group setting as a way of making the objects studied both interested in the study and objecting to the study.This argument has a very great debt to the active interview and the numbers of other central social construction perspectives to interviews.But whereas the argument of the social construction approach emphasizes the social situation between parts in the interview, I suggest focusing on the construction of the scientific setting.

www.intechopen.com
Natural Interactions in Artificial Situations: Focus Groups as an Active Social Experiment 63 the focus group that we can produce these new versions.The focus group should not only be approached as a setting for investigating what we expect, but as a tool to add further courses to an explanation of a phenomenon.By epistemologically viewing findings as versions we, as researchers, could focus less on biases and focus more on how the focus group could be a tool to investigate versions of the phenomenon that we would probably not see otherwise.
In particular, two aspects in the active use of experiments should be considered: (1) how the researcher -intentionally or accidentally -creates a specifically artificial situation through composition of groups and facilitating interaction, and (2) how this can be used productively in making interesting data.In the following I will address these two aspects separately with an empirical example.

Composition of focus groups
One of the strengths of the artificially created social setting is that it enables the researcher to actively use the different positions of the participants to influence the social situations.This requires the researcher to consider the sampling of participants and group composition thoroughly.The methodological concept of sampling is primarily used in order to be able to strengthen the validity of general results in special quantitative data.When sampling is used in relation to focus group studies it in some cases relates to a practice of screening possible participants in order to ensure a combination of group participants in the same way as when dealing with quantitative studies (cf.Flick, 2007).That way it becomes possible to argue for general findings based on a quantitative notion of validity.However, I will argue that it is essential to look at the composition of groups for another reason as well.It is argued that composition of groups is not only a matter of "sampling" persons with a certain biography "outside" the group but also in question of how these persons can be made to interact inside the group.
The group composition is crucial in order to construct a social setting which enables the participants to interact in unexpected ways.In the research project on teenagers and alcohol I used a variety of group compositions which provided a unique possibility to explore the making of different versions of the phenomenon of teenage drinking.Even though the intended strategy of the focus groups aimed to facilitate specific aspects of the phenomena, it was not always these versions that appeared in the material.Thus some might argue that the strategies did not work.However they worked out well in terms of making it possible to discover how the different compositions of the focus group actually produced unexpected knowledge of the phenomenon.Here the research design showed -as it probably will in many studies -a dual practice of on the one hand trying to actually understand how data was formed by letting the data production strategy be part of the analysis.And on the other hand letting the experiments be as open as possible in order for the researched to be able to object and react with productive interactions.
In the focus groups sampled from the school classes, we decided to conduct male, female and gender-mixed groups. 5This configuration was chosen primarily in order to make the girls more comfortable with talking about alcohol.We assumed that the girls would be a bit shy when interviewed in gender-mixed groups, because we supposed that males would drink more and would have a tendency to talk more enthusiastically about their binge drinking.In order to meet our strategy of making the girls more comfortable, the girls only groups were moderated by a female moderator.Surprisingly, the expected data was not produced by this sampling strategy.It soon became clear that the girls in the 8 th grade often had more experience with alcohol than the boys their age, because some of them had been partying with older boys.In the gender-mixed groups the girls did not allow this 'male dominance' to silence them.Moreover, in a few of the boys only groups very few had alcohol experience, and the expected boastful behavior around alcohol was less evident (I will return to this).Thus, the sampling strategy might be considered a failure, because the construction of groups did not produce the expected data.It was, however, only because of the sampling strategy that this was revealed and could be included as a finding.My discovery about the girls in particular implied that I changed moderation and group composition strategy so that the groups of both boys and girls were moderated by both male and female moderators and that implied a greater openness towards the possible outcomes.In consequence, I changed the composition of the groups to be more in line with Despret's reflection on experiments -as active experiments that are supposed to make interesting and surprising data, and in which both the interviewed and the researcher get themselves involved.
In the present study the same teenagers were interviewed several times over the three succeeding years (8 th , 9 th and 10 th grades).This longitudinal design offered a good opportunity for investigating how the experienced drinkers positioned themselves in relation to the less experienced drinkers in the early grades, and how the different subject positions were negotiated when the participants changed their behavior.The experimental group composition also made it possible to see how the heavy drinkers reacted when some of the non-drinkers later took up their binge-drinking lifestyle. 6By the 9 th and especially the 10 th grade interviews most of the participants were partying and binge drinking on a regular (weekend) basis.Thus that the distinctions made by the experienced drinkers were no longer considered the most important by the other participants.One might suspect that the experienced drinkers would drink even more in order to keep their positions as the more mature, but in general this was not what happened.Instead the more experienced drinkers downplayed the significance of drinking heavily and in this way tried to construct those who had just started partying and drinking in the 9 th or 10 th grade as beginners or amateurs who had not yet found the proper way to practice drinking.However, most of the late starters were not willing to accept this social identity.Instead, they related to the social understanding that the experienced drinkers were able to make powerful in the groups (e.g. 6 This study used a notion of procedural informed consent (Heath et al., 2009).Even though the teenagers involved in the study were fully informed of the study before partaking in it at grade 8 th , we would not have been able to actually in detail describe the process of research all to the end of the longitudinal range.As such we tried constantly to inform and to accept that some might chose not to continue their participation.In this respect it was interesting that some of the very empowered participants (girls with intense drinking experience in the grade 8 th ) seemed unwilling to partake in the grade 10 th .Their objection could be seen as a central finding; they were no longer able to set the agenda and would thus rather omit to partake.
by presenting e.g.vomiting at parties as childish).They tried to re-present their drinking practice as more mature (according to the more experienced drinker' perspective).Some of the late starters did accept the image of the beginner, but used it constructively to change the presentation of themselves from a person who used to be quiet and non-partying but has matured and is now outspoken and sexually active (Demant, 2009).
As we have seen, the focus group creates different results and knowledge, depending on how we choose to compose the groups.The different versions of the phenomenon can be viewed as the strength of the method.The focus groups turned out to be a kind of social experiment; the alcohol discussions formed a specific social setting that let the researcher look into the more refined aspects of how alcohol works as a symbolic marker among youths.The experiment enables the researcher to construct different analytical perspectives on the same topic -through forming of different forms of focus group settings and thereby pose unpleasant, yet productive, questions to his or her own conclusions.It could be argued that the focus group data on the early debuting teenagers as well as the late starters might be possible to observe through ethnography in school classes.However, I believe that these findings were possible only because the teenagers were put into the experiment of focus groups and confronted with questions on drinking and that we -the researchers, as well as the teenagers themselves -were able to see different versions of realities.In the groups we were able to observe how teenagers both conformed and objected to symbolic dominance as well as how some worked hard to maintain their symbolical dominance through their drinking practice.Had we instead choosen a methodology of ethnography in school class-rooms we might not have observed the power relations around alcohol.In the class-room a separation of nondrinkers and drinkers may have made it impossible to observe the practice of dominance that is related to the alcohol-party identities.Thus, the "laboratory" of the focus group setting lets researchers target and challenge some of the 'natural' or matter-of-course practices which exist in the interviews with the teenagers and in this way gives the researcher a privileged insight into to the phenomenon.

Interaction and moderator
The interview guide creates an underlying structure for the focus group interviews.Relevant statements relate to the topics of the interview guide and are in this way possible within the discourse of the interview.In the study of youth and alcohol I, as a moderator, tried to build up an ambience in which the participants did not understand the interview (themes, style etc.) as an expression of a moral condemnation of their lives.In line with this, I tried to avoid facilitating a health or risk discourse at the outset of the interview.It is, however, open for discussion whether this has in fact made the group discussions more "naturally", or if it has facilitated another discourse.I will now address how moderation and themes can play an active part in opening up the interview.In particular, I will focus on the concept of natural data and the productiveness of failed interview situations where interview participants objected to the researcher's projection of them.Green & Hart (1998) argue that a focus group should have a kind of natural ambience.They argue that the participants' ability to act naturally in relation to each other will make it possible to have sincere and honest discussions, especially when interviewing children and young people.In their discussion of how to interview school-age children, they argue that the validity of the group results should be valued in relation to the success of creating a natural setting.They consider the setting to be more natural if some kind of chaos is generated in the focus group thus resulting in a natural ambience.7These reflections are indeed very relevant and are visible in my own moderation technique of producing a social space for discussion close to the participant's everyday lives and interactions.However, as discussed, the focus group never puts the participants in a natural setting: even though aspects of their daily lives are discussed, they would in most cases be put on display in ways that are new to the participants (Warr, 2005), and the social situation forces the participants to engage in such a discussion.To discuss this, I will focus on what happened when I used bottles (alcohol, beer, alcopops and energy drink bottles) as stimulus material in the 8 th grade groups (14 years of age).The bottles were put on the table half an hour into the interviews when the participants had discussed the leisure life in their home town.The stimuli were introduced as a way of shifting focus towards alcohol and drinking. 8n most of the groups, the following happened: when the bottles were put on the table, it caused a lot of noise and inspired the participants to point comment and cheer. 9The ambience changed radically into a form of effervesce.Sometimes the participants asked for permission to grab the bottles, at other times they just grabbed them and started telling stories about drinking or about their attitudes towards the different kinds of alcohol.The alcohol products were used as a stepping stone to defining themselves in relation to each other.The bottles with alcohol seem to slide into most of the groups with some kind of naturalness.A quote from a group of girls from a 9 th grade focus group seems to confirm this: "You sit and talk like right now [in the focus group], just with a couple of beers on the table.Well, that's how I think of a house party."This statement could be understood as if the focus group discussion were very close to the situation of a group of friends talking together -a natural situation.Due to this, it was a great surprise that particularly two focus groups with boys did not respond when the bottles were put on the table.When the moderator asked about the products, it turned out that the participants knew of them and had tried drinking them in small amounts.However, alcohol was not a central element of their leisure lifeworld.As the bottles stood there on the table, they did not relate to the participants' stories or in any way interfere with their reality.The bottles were present in the discussions but did not have any central symbolic effect.The reluctance, uneasiness and negative comments to a certain theme or in a focus group as a whole are known from other studies (Pösö et al., 2008) but are seldom reported (Michael, 2004b).On one side, it would have been interesting to follow what other aspects that could have been relevant to them.Efforts in order to interest the boys in the research topic with another designing of the social experiment (other stimuli and other questions) could have contributed to the study.On the other side, it is also an interesting finding in it-self that the boys silently refused to participate in the discussion.The research 'objects' objected to use Latours (2004) words.The boys used their collective power to obstruct the dominant discourse facilitated by the moderator (through choice of topics and stimuli).This was a very strong and coherent local counter discourse that was in opposition to and awkward for the moderator.At first I considered these groups as failed.In order to keep the discussion going I found myself (as a moderator) asking more and more direct questions to the singular individuals in the groups, and the focus group changed into a group interview with a structured "guide".And no discussion arose as in the other groups.The groups were, at first, not transcribed as it seemed pointless to use them in the analysis.Even though these focus groups were considered failed, these groups nevertheless made it possible for the participating boys to position themselves rather powerfully (in this particular setting) by actually objecting to the maturity discourse, which I, as a moderator, was facilitating.Stengers (2000) and Latour (2005) discuss how humans often become docile and obedient "objects" when studied by social scientists.However, the incident of the failed focus groups proves to be successful.The method was able to produce a situation where the "objects" would not only confirm or reject a thesis, but where some of the versions of the practices were put into words.This is not only an issue of visualizing seldom discussed aspects of a phenomenon.The collective power that was produced in these specific groups -these social experiments -made the focus group a central tool in producing interesting data that were new to the researchers.The objections would probably not have existed in casual class-room interactions between pupils but were produced in the social experiment of the focus groups.Michael (2004aMichael ( , 2004b) ) argues that a somehow failed or mistaken element of the data collection is actually rather productive.In the present focus group study I similarly found that the way the participants objected to the (unspoken) expectations of the focus group actually provided what turned out to be very different findings.The failed interviews made it clear that the use of alcohol bottles as stimuli was thus not transparent (neutral), but worked to facilitate one particular kind of group discussion; a powerful discourse which connected alcohol experience to both maturity and gender, in ways that made it very hard for those with relatively little alcohol experience to reject it.An objection to this dominant way of self-presentation would position those in opposition as both childish and not presenting a self that would be culturally understandable in the setting of the Danish teenagers intensive drinking culture (Demant, 2009;Demant, 2007;Demant & Järvinen, 2006).Thus, in the end these "failed" interviews proved to be very central because the versions that they actually provided became central contrasts to the other (dominant) versions of the phenomenon.
As we have seen, the focus group study is anything but a modest observation of facts (Haraway, 1997).It is an active co-production of reality by way of moderation, topics and sampling (Whatmore, 2003).Any moderation technique is active in some ways.When we acknowledge this, it first and foremost makes it more valid to work directly with asking questions that counteract the discourse in the interviews or in other ways interact in the social setting of the groups.The examples presented show how these aspects of the active un-naturally setting would not have occurred by themselves, and how they actually contributed with central versions of teenage drinking.And secondly, the examples also point towards the fact that the experiments should not alone be acknowledged as a control of the social focus group setting, but that they might as well open the study to constructive mistakes.

Conclusion
I have argued that if researchers explore the artificiality of the focus group setting we can provide very good and specific data on social interaction.It also enables us to transcend the unproductive discussion of when focus groups are biased in relation to real life situations.The un-naturally situations of the group are exactly the inscription devices that are able to articulate some of the social relations of which the vast majority is rather mute in other materials.In comparison, ethnography enables researchers to observe social interactions and relations in a natural setting.This is unquestionably the reason why it is often in the ethnographic field work that we sense some of the interesting -and hidden -logics of the social.However, even hanging out for a very long time in a "street corner society" will unlikely provide inscriptions of social interactions at the level of focus group interviews.This is because the focus group method is a device that can both transform the interaction in the present into an easily accessible form of data for analysis, and to an even greater extent because the artificial setting actively makes silent versions of the social interactions appear.As such, considering the focus group as an experiment facilitates an inclusion of some of the social logics that often remain mute in other kinds of data.Focus group's social experimental character tend to produce data that would not have been found or produced in natural occurring data.As such, these data are constructed in the specific setting of the focus groups.I have argued that this data could be understood as versions.Desprets concept of versions is closely related to Donna Haraway's (1988) concept of situated knowledge.Both authors focus less on the epistemological problems with relative data, but argue that the concept of situated knowledge or versions is an opportunity to integrate science production as a part visualizing the many different and conflicting aspects of the phenomenon studied.Latour argues in the same line, when he describes the fact construction of science as a process of human and non-human actors that interact with the studied and how facts become possible only in the different inscription devices that are involved in translating them.As such different facts will be possible because science makes them solid through its construction."Construction" has different implications from those of "social construction".Results do not become less real when we accept that they are a result of a number of inscription devices like focus groups.Instead we will have to be critical towards these inscription devices and ask the question: Have we, as researchers, made productive interactions with the objects studied so that these have had the possibility to act?Methodologies using social or psychological experiments draw attention to the question of ethics.By applying the metaphor of the social experiment and being skeptical to the notion of the natural data in the focus group I also wish to expose the method of focus groups more to an ethical discussion.The process in focus groups may expose and in some cases suppress the participants.Thus the focus group always runs a risk of being unethical.Even though these issues become the more highlighted in relation to using focus groups as more active experiments they are as important in more passive forms of focus groups.It is the intention that the use of the concept of experiments in describing focus groups should draw more attention to these ethical problems and help push for a development of the discussion of ethics in relation to the use of focus groups.Even though focus groups are able to provide interesting data, it is necessary to discuss in what cases it would be more ethically sensitive to use other methods, and how to deal responsibly with some of the sensitive issues when they appear in -often unexpectedsituations.However, it also becomes an issue to discuss how social empowerment can happen in focus groups when participants object to being objects of research in a specific framing, thus making focus groups more ethical?!Even though researchers design focus group studies in order to actively use the social setting experimentally for producing data, we should not -and cannot -control data.This is paradoxical: On the one hand, researchers are strategically trying to produce certain versions of reality through sampling, moderating etc., and on the other hand anticipate that these experiments will not necessarily produce the kind of data we expect.However, an active use of the experimental designing makes researchers focus on the fact that we are actually producing situated data.This means that we cannot hide behind a modest veil of passive methodological designs, even if we want to.And, secondly, the use of the active experiment would make researchers focus on how to produce settings that would unveil new versions of reality.Thirdly, these experimental designs should also make researchers take interest in what happens when things go wrong: what aspects of the design actually enforced these situations and ask what are the logics of group participants objecting to the data design that the researcher made (either manifest or latently).
The reintroduction of the social experiment in relation to focus groups can make researchers more aware of how the method can be active in data production.The epistemological position of the social experiment as an active inscription device is as such not (only) a critical view on focus group methodology.It is critical towards the argument that validity is obtained through a "close to natural" situation.However, it does not abandon focus groups with a potential relativistic status as has happened within some social constructivist approaches.Instead it is argued that focus groups as social experiments situates findings within a more clear and transparent methodology.The validity of focus group methods must, within this perspective, be judged on its ability to provide (and clearly describe) active experiments that can contribute with situations for groups participants to provide other versions of the studied.These kinds of approach are not at all absent in most focus group studies.But reframing the focus group epistemological as social experiments forces us as researchers to pinpoint how the focus groups can be a productive social space for making translations that are less possible to follow in other data materials.