Stepping Beyond Transcripts: A Framework for Analyzing Interaction in Focus Groups

ABSTRACT Interaction is a much-claimed attribute of focus group research yet is often deficient in analysis when its essence can become lost. In this paper we aim to develop a flexible framework that can be operationalized and replicated when attempting to encourage and capture interaction. Working as outsider researchers with John Lewis & Partners (the UK’s largest employee-owned business, with 80,000 employees known as ‘Partners’), we conducted eight focus groups asking 18 questions about the company’s giving activities and associated decision making. Using the transcriptions, we analyzed interaction through a taxonomy of questions, laughs, and pauses, identified as the features of both interaction, and sequential interaction. Employing a two-stage approach for encouraging, capturing, and evidencing interaction, we developed an exploratory framework. Through a transparent audit trail, we reduced the data to points of impact. We propose that these present a meaningful starting position for the theorizing iteration of the data.


Introduction
The purpose of this paper is to advance interaction analysis in focus group research.We explore how interaction benefits (which we believe are significant) can be operationalized into a framework that can be replicated by practitioners and researchers.
Commencing qualitative analysis with the writing up individuals' views, researchers struggle to highlight how the interaction happens and its significance.This neglect can lead to flaws in acknowledging interaction, failure to discuss how interaction can be analyzed, and, arguably, incomplete analysis.Useful data is effectively ignored, yet interaction is both a key characteristic and a benefit of focus groups, distinguishing focus group methodology from other methods (Kitzinger, 1994(Kitzinger, , 1995;;Smithson, 2000).We encourage researchers to look at fresh ways to advance and capture the conversations between participants within a focus group.
Without an attempt at interaction, analysis in focus group research design lacks completeness.Interaction analysis provides an advanced appreciation of the conversations that take place within a focus group.We theorize that there is an unexplored liminal zone within interaction analysis.Synthesizing transcripts with interaction analysis can lead to the identification of the questions that create the most impact and new levels of theorizing.We present an exploratory framework of interaction analysis in focus groups, working with a real-life organization and in so doing, we demonstrate how interaction can be conducted, applied, and presented.
We propose a two-stage framework.Stage One creates the conditions to encourage and capture interaction, from the initial planning of the focus group to the researcher extraction.Stage Two illustrates the value of interaction analysis through the employment of two iterations.We test this approach using a real-life organization, namely, John Lewis & Partners, the UK's largest employeeowned business.Famous for its ownership structure, it employs over 80,000 people, all known as 'Partners' (John Lewis Partnership, 2022).As outsider researchers, we conducted eight focus groups (comprising 52 participants), investigating the philanthropic decision-making process that was used within the organization.
Our research aims to increase the facilitation, analysis, and discussion of interaction within focus groups.There are several stages involved in this process.Initially we discuss ways of encouraging interaction, then outline possible ways to both record and analyze interaction.We also articulate why interaction is important with the associated aim of raising the profile of interaction within the focus group literature.

Development of Focus Groups as a Research Technique
Focus group research as a method matured and evolved outside the inductive tradition (Basch, 1987) and was originally developed for assessing radio audience reactions.Merton (1987), who is often credited with the development of focus groups, critiqued Lazarsfeld (1975) for guiding responses and not elucidating spontaneous expressions from participants and said that the discipline evolved because of a publishing misnomer between the uses of the terms 'focus group' and 'focused interview with a group'.
The journal Qualitative Sociology introduced Morgan and Spanish's focus group work as a relatively new research tool (1984, p. 253), thereby no longer referring to Merton (1967Merton ( , 1987) ) as the original source of ideas.Morgan (1996) describes focus groups as a research technique for the purpose of collecting data through group interaction on a topic predetermined by the researcher.He further identifies the source of the data as being collected through interaction, with the researcher's active role being to facilitate group discussion for data collection purposes.The main purpose of focus group research as proposed by Gibbs (1997) is to draw upon respondents' attitudes, feelings, beliefs, experiences, and reactions in a way that would not be feasible using other methods (such as observation, one-to-one interviewing, and questionnaire surveys).

The Outline of Our Challenge
The benefits of focus groups should not be underestimated.A major advantage of such research is the opportunity to observe participants engaging in conversation without the moderator leading the conversation.Focus groups can be constructed as a social occasion in which participants are encouraged to express what they know, making a valued contribution through an opportunity to share views in a positive environment while interacting with others.Wibeck et al. (2007) consider interaction to be a hallmark of focus groups, yet it is seldom evaluated, often ignored in analysis, and rarely discussed in empirical research.This may be because researchers either do not know how to analyze interaction or simply fail to consider the possibility in their approach to the work.Facilitation strategies need to be clear and consistent (Rinkus et al., 2021), particularly between groups.Gasiorek et al. (2021) articulate that combining approaches into a single framework is not simple.A key area is the consideration of how to facilitate the interaction within the group, and we argue that that is where the interesting data arises.Platt (1992) advises that the starting point should be a logic of design.Morgan (2010) introduces an ambitious three-stage agenda for improving our understanding of the role that interaction plays.First, calling for detailed investigations into how interaction can be conducted (operationalized in our research as Stage One); second, for a reflexive examination of the value that is added by analyzing interaction; and third, for effective presentation of results, with consideration being given as to why interaction insights are beneficial (both operationalized as Stage Two).
To do this a logic of design is introduced that is flexible and replicable.The two stages encourage, capture and evidence interaction, forming the basis of the methodological discussion, while the application of the stages forms our theorizing.Multi-levels of interaction analysis are applied to the data, involving sequential interaction and 'quasi-statistics.' Becker (1970) introduced quasistatistics as non-precise counts obtained from inductive research.This results in a novel and replicable approach, that is tested on empirical data, and permits advances in debate, leading to possible theoretical development and practical insights beyond the bare transcripts.
Our paper proceeds as follows: first, we review the use of a taxonomy of questions and face-work to create the conditions for interaction.Second, we describe our two-stage framework, which provides a logical process to encourage, capture, evidence, and apply interaction.In the subsequent discussion, the value of interaction analysis is highlighted both by considering reactions to each question (as measured by quasi-statistics) and by noting the number of contributing participants.We illustrate our framework by using the case of John Lewis & Partners and then consider the practical relevance of the method we propose.The paper ends by acknowledging the limitations of our method, presenting further avenues for research, and offering our concluding thoughts.

Literature
Without the identification of interaction and its analysis, valuable data is omitted, differences between groups are not captured and questions that create impact may be overlooked.Thus, capturing interaction helps identify issues of importance which arise during focus groups.Rinkus et al. (2021) state there is surprisingly little said about how to generate verbal interactions yet most primary data analysis focuses on verbal content.Little attention has been paid to analyzing participant interactions and how knowledge might be socially constructed.The failure to acknowledge and quote sequential interaction, referred to as the 'conversation between participants' (Kitzinger, 1994, p. 104) and 'sequences of group interaction' (Grønkjaer et al., 2011, p. 17), devalues the very interaction that produced the data in the first instance (Morgan, 2010;Wibeck et al., 2007).Stewart and Shamdasani (2014) and Fern (1982) argue that stimulating interactions among group participants generates more information than individual interviews.

The language of hint and face-work
In focus group research, social interaction sets conditions for higher levels of participant interaction, awareness, and analysis.Glesne and Peshkin (1992) suggest that moderators should adopt a learner role in their relationship to focus group participants.Goffman (1955) proposes that moderators should use a 'language of hint,' including well-placed pauses, and specifies the following of a specific line of questioning in order to maintain momentum.By not adopting a learner position, the moderator may confuse participants, 'fall out of line' with questions, and skew results.Participants may (incorrectly) perceive there to be a right answer which the moderator is eagerly awaiting.Goffman supports the maintenance of face as a condition for interaction.'Face-work' is 'the actions taken by a person to make whatever he (she) is doing consistent with the face' (Goffman, 1955, p. 226).Put simply, by adopting a careful approach to 'face work' the moderator can facilitate both interaction and encourage sequential interaction between participants.In so doing, a greater depth of discussion occurs.Asbury (1995) affirms that group interaction can yield richer information than individual interviews with the same participants.
The real value, however, comes from the researchers' exposure to the participants' own thoughts as the means of self-expression.Conversation between participants (which we term sequential interaction) highlights differing worldviews about the particular situation being explored.Types of communication such as pauses or laughter (referred to as features of interaction) can also be recorded and presented (Becker, 1970;Lazarsfeld, 1935).There is also merit in making connections between a planned taxonomy of questions and sequential interaction between participants.Finally, the planning of the questions is relevant: locating higher-order questions in the center of the question set can facilitate better understanding and discussion (Bloom, 1956).Goffman (1955) contends that focus groups are understood as social enactments, inspiring interaction analysis.Notwithstanding this, analysis sections of focus group data within research papers often fail to expand on how to capture and analyze interaction (Duggleby, 2005;Halkier, 2010).Eisenhardt (1989) discusses a common lack of clarity and few exemplars to guide research, including forms of analysis.A particular issue is that many articles and academic research projects appear to treat the data identically to data obtained from individual interviews (Wilkinson, 1998).Kitzinger (1994) reviewed over 40 published reports of focus group studies and found that not one reported upon the conversation between participants.Cyr (2016) analyzed focus group literature within political science and sociology articles over a ten-year period (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013) to see how often the social nature of interaction was captured and estimates that only 0.69% of articles in political science and 1.2% in sociology even mention focus groups, which is noted as surprising, given the resurgence in the data collection method.

Motivation for Interaction Analysis and Recording Interaction
The methodological rationale for doing this type of interaction analysis is to enable both more detail to be captured and valuable data uncovered.This aids qualitative analysis and subsequent theorizing.There are two areas to consider for the development for encouraging and capturing interaction and the gathering of relevant evidence: sequential interaction and quasi-statistics.

Sequential interaction
This involves developing the participants' conversation to be presented in the form of counts, allowing for interaction outside the structure of researcher-constructed questions.Content found in sequential interaction may feed into the formulation of new hypotheses and fresh insights (Bratton & Liatto-Katundu, 1994), leading to unexpected findings, new research questions, and new hypotheses.Measuring sequential interaction is important because it helps researchers identify what is important to the participants, rather than what the researcher believes to be so.

Quasi-statistics
The second area is the introduction and employment of quasi-statistics, a process which involves collecting descriptive counts of interactions that take place in a focus group.Barton (1995) and Lazarsfeld (1935) both claim that the inclusion of frequency data presented as quasi-statistics separates the data into categories (Onwuegbuzie et al., 2009;Warr, 2005).In addition to sequential interaction the counts of categories can include laughter, pauses, and crying, developing a logical structure to assess categories that are predetermined by the researcher in a focus group setting.Barton and Lazarsfeld (1961) advise that the analysis of quasi-statistical data can be made through a logical structure to provide direction for the inductive researcher.Becker (1970) suggests that quasi-statistics provide useful information that is fundamental to inductive data (Maxwell, 2012;Morse, 2003;Onwuegbuzie et al., 2009;Onwuegbuzie & Teddlie, 2003;Warr, 2005).Although insightful, these approaches lack a means to operationalize interaction into achievable steps that researchers and practitioners can engage with.The starting point, therefore, for a framework of a logical method design framework is how to create the conditions to capture sequential interaction (and other related features).

Method
Identifying Incompleteness in the Current Method Pan and Tan (2011) argue that, because interaction cannot be prescribed, the methods for documenting it are not easily translatable into specific steps.Yet a discipline without exemplars is an ineffective one (Kuhn, 1981), so there remains the potential for developing new and flexible frameworks that can develop the discussion on how to analyze interaction in focus groups.
We ask the question: how can analysis of focus groups take account of, and present for scrutiny, the interactive nature of the data?Blending Goffman's (1955) face-work and Barton's (1995), Lazarsfeld's (1935), Becker (1970), and Warr's (2005) quasi-statistics into our exploratory design, we provide a specific response to Morgan's (2010) ambitious three-stage agenda by detailing our investigations, examining their value, and presenting the results together with a supporting narrative discussion of why these are beneficial.The research employs an exploratory design within a two-stage approach in order to encourage, capture and evidence interaction.The application forms the results section.

Participants and setting
As a method, focus groups offer a framework such that the findings are not typically generalizable.Replication is not possible as each group differs in its constitution.Nevertheless, the research approach can be replicated and can follow certain parameters.Discretion is required regarding voices that may dominate the discussion; similarly attempts need to be made to engage the taciturn or reluctant participant.Focus group settings do not conform to an ideal and should be flexible enough to adapt to the environment.There is no such thing as a neutral or ideal setting; Fern (1982) points out that the setting can play a significant part in influencing the discussion.
The researcher is accountable for their own intuitive choices around transcription; there may be dilemmas when all participants speak together/over each other.By planning and encouraging interaction this dilemma is amplified, in which case the researcher will need to decide on how to record and document this.

Location of Data Collection
In order to describe the method, our own study was carried out following the same process in each focus group.We collected data from eight such groups, involving 52 self-selected participants.The participants came from two United Kingdom settings -five groups from Newcastle upon Tyne and three in Edinburgh -both from branches of John Lewis & Partners.As outsider researchers we obtained full ethical consent.Each focus group was coded A to H, noting the location, Partner responsibility, and number of participants in each focus group

Two-Stage Framework
One way of adding to methodological rigor is to subject data to a number of iterations (Rothwell et al., 2016).Pratt (2009) suggests that the narrative be supplemented with plenty of data summary devices.Our interaction analysis comprised two stages, replicated in all eight of our focus groups.

Stage One -Encouraging and Capturing Interaction
Stage One creates the conditions for encouraging and capturing interaction and consists of four steps: (1) Step One covers the development of a taxonomy of questions designed to encourage interaction.
(3) Step Three examines interaction in action by asking research questions planned within a taxonomy.(4) Step Four considers researcher extraction and departure from the interaction (Goffman, 1955), referred to as 'cool-down.' We now explore these steps in turn.
Stage One -Step One: taxonomy of questions.Prior to the group commencing, semi-structured focus group questions were constructed, drawn from the literature, and ordered within a taxonomy (Bloom, 1956).These questions [Appendix 1] were designed in such a way as to stimulate both conversation and interaction between participants.They began with how, who, and process-seeking questions (encouraging debate), before climbing to higher-order questions, categorized as evaluative questions seeking internal evidence (Tarman & Kuran, 2015).The question level descended toward the end, with lighter, knowledge-based questions bringing the focus group to a close.As Goffman (1955) claims, the maintenance of face is a condition of interaction.The moderator aimed to say little, giving eye contact and encouraging nods as a replacement for narrative.
Stage One -Step Two: researcher integration and warm-up.The purpose of this step was to make participants feel comfortable with the moderator being present, which is crucial in developing interaction with participants.Visits were undertaken before the focus group meeting.Early on in the research a decision was made to mirror how participants dressed (Goffman, 1959., Kitzinger, 1995), providing a further opportunity to integrate and make participants feel at ease.Broom (2005) advises that the first personal contact need not be rushed, suggesting that 15 minutes should be invested in acclimatizing the group.Rinkus et al. (2021) recommend spending time on introductions.This time is used in ensuring that participants fully understand the purpose of the focus group in which they are participating.Goffman (1955) concurs with Broom by stating that participation with others leads to commitment and therefore interaction.
We used 'face-work' (Goffman, 1955) with a smile, direct eye contact, and a light touch.Each focus group participant was personally welcomed by the moderator.Participants' names were not featured, to preserve anonymity, as ethically agreed in advance because of the nature of the discussions in this context.Serving refreshments provided integration, creating a relaxed atmosphere for social interaction.Our intention was to encourage participants to talk freely, to interact with the moderator, and, in particular, to commence conversations with each other (sequential interaction).
Once we perceived that everyone felt at ease and comfortable, following Goffman's advice, the purpose of the focus group was explained.
Stage One -Step Three: interaction in action.Positioning the moderator slightly apart from the focus group allowed the interaction to be transferred from moderator-participant to participant-participant capturing the conversation that takes place between participants.We followed the advice of Kitzinger (1994Kitzinger ( , 1995)), who suggests that the moderator should take a back seat at first, accommodating a type of structured eavesdropping.Similarly, we adopted a learner role (Glesne & Peshkin, 1992), with a style offering no emotion or encouragement that could influence data.
Stage One -Step Four: researcher extraction and cool-down.The cool-down and extraction stage (Goffman, 1955) involved a transfer from participant-participant sequential interaction back to moderator-participant.Each participant was personally thanked, replicating the process applied in Step Two, including for example, a handshake.
Stage Two -Evidencing Interaction.The approach we adopted to record evidence involves the collection of quasi-statistics.This comprises two separate iterations -quasi-statistics by focus group and quasi-statistics by question -with each iteration consisting of three steps.Stage Two centers on the analysis of interaction data captured in Stage One.
Stage Two -Iteration One: quasi-statistics by focus group.This consists of three steps: (1) The number of sequential interactions per group.
(2) The average number of sequential interactions per question and per group.
(3) The development of researcher-defined categories.
In Step One, the number of sequential interactions per group refers to the conversations that take place between participants (Bratton & Liatto-Katundu, 1994;Kitzinger, 1994).The count stops at the point of the next moderator intervention (be that asking the next question or to prompt debate).Step Two refers to the average number of sequential interactions between focus groups.Connecting the results from Step One to Step Two permits the researcher to see what questions prompt patterns of sequential interaction to emerge.Finally, Step Three allows further interaction categories to be added by the researcher.In our research we added features of interaction such as laughs and pauses; pauses, for example, could indicate a higher level of thinking (Albergaria-Almeida, 2010; Wilen & Clegg, 1986) prior to offering a question response.The results of all three steps shed light upon the levels of sequential interaction, and features of interaction which may differ across groups, thereby facilitating comparison [Table 1].We acknowledge that by allowing participants to direct the conversation, different groups may focus on different topics (see the discussion in Rinkus et al., 2021).Where the purpose is to compare data across the groups, the moderator will need to take a more active role in managing the conversation direction.
Stage Two -Iteration Two: quasi-statistics by question.This consists of three steps: (1) The number of contributing participants.
(2) The range of contributing participants measured against the number of focus group attendees.
(3) Employing the data from steps 2 and 3 leads to the identification of the sequential impact questions.
In Step One of Stage One we counted the number of participants who created the sequential interaction in responding to a particular question.In Step Two, we looked at the range of contributing participants, measured against the number of attendees [Table 2].Connecting the results from Steps Two and Three allowed us to explore the identification of the sequential impact questions.
Step Three was not in our original method design logic and was added later, as we realized that quasi-statistics by focus group without consideration of quasi-statistics by contributing participants was less insightful.To only map sequential interaction conversations between participants means that there is no recognition of the range of contributors, meaning that if only a few participants were to make up the sequential interaction the count would become misleading.By introducing the quasi-statistical counts of contributing participants, we can see a clear pattern between the sequential interaction and the range of contributing participants, reflecting what prompted these reactions [Table 3].The two iterations presenting quasi-statistics by focus group and by question are achievable, clear, replicable, and transferrable [Figure 1].

Method Summary
Focus group research does not assume ideal conditions.We allow for this by focusing on what can be done with the data in order to create an exploratory framework, we posit the two-stage design as our exploratory framework for sequential interaction analysis.The two-stage method measures interaction analysis and was manageable across the eight focus groups in this research, dealing with 2230 sequential interactions and 509 counts of participants contributing to sequential interaction.More researcher-defined features of interaction categories could have been introduced: for example, crying and body language.Too many categories of interactions would be cumbersome if employing multiple focus groups.
We decided to focus on the achievable, manageable, and replicable, with the two-stage approach offering a fresh view.It is designed to be flexible, representing a transparent process for capturing and evidencing sequential interaction, resulting in the identification of the questions that created the most impact for each focus group.The application of sequential interaction is dealt with in the Findings discussion.Responding to Morgan's (2010) ambitious three-stage agenda, we aim to use these data to improve the understanding of the role that interaction plays.

Results 1
Although there is agreement around the benefits of interaction, the issue we have explored is how to operationalize sequential interaction within focus groups into an exploratory framework that can be utilized by practitioners and researchers.By listing possible stages and iterations, we have provided a base for other researchers.However, the research is not without its issues.Even in our own study we could not necessarily identify all aspects of sequential interaction.For example, in focus group F all participants were talking together, and we were unable to count how many participants contributed to the sequential conversation.

Capturing Features of Interaction
Stage One was specifically designed to create the conditions for and encourage sequential interaction between participants.It was important to capture how each of the 18 research questions engaged the participants.It was useful to capture the features of interaction, for example, groups A and F laughed the most, C the least; A also paused the most, while F and H did not pause at all when answering questions.Our findings offer deeper meanings, requiring us to synthesize the results with the number of contributing participants, which we do later in this paper.
Making comparisons by question mirrored our planned question taxonomy.Question 10 created the most pauses (see Appendix 1), was ordered as a high-level question, and was placed in the center of the taxonomy.It required participants to think about sharing examples, so the pauses might have indicated that participants were thinking about which examples to select and link to the benefit of giving.In contrast, Question 1 was designed as an introductory question based on knowledge of the organization's giving activities, and it generated the highest number of laughs.A possible reason for this is because the responses generally related to storytelling.Early questions were designed to encourage sequential interaction and other responses such as laughing.
In the groups conducted we found that the sense of social occasion and learning outcomes aligned with benefits reported in the literature.For example, Bristol and Fern (1996) claim that there is evidence that participants in focus groups find the experience stimulating.Sequential interaction captures the knowledge transfer between participants in the form of conversation, further advancing reported benefits.At the cool-down stage, several participants registered their experience of focus group benefits, providing support for these observations: 'I was nervous at first, but it was a relaxed, fun and comfortable session.' 'Really helped to remind me just how special we are.' 'This has been a great thing to be part of.' 'Excellent session, very engaging.I thought I was aware of all that went on and was wrong.' 'I actually realized I knew more than I thought and learnt a little more, so it was beneficial.' 2 This process resulted in the development of new knowledge among the participants and an awareness (on the part of the moderator) of which particular questions prompted pauses, laughter, and tears.Although initially unexpected, we recorded 17 occasions of tears throughout the eight focus groups.

Evidencing Sequential Interaction
In Stage Two, we addressed quasi-statistics by both focus group and question, identifying the number of sequential interactions (conversations) between participants, and the interactive features that occurred.The number of sequential interactions alone could be misleading without reference to the number of participants.For example, in a focus group of ten, if only two participants contribute to the conversation and the other eight are silent, this is different from all ten taking part in the sequential interaction.Quasi-statistics need to account for this and record the range of contributors in order to obtain full meaning from the data.Results are more meaningful when the sequential interaction is reported in conjunction with the relative number (or percentage) of contributors.For example, the questions identified as having the highest levels of sequential interaction did not have the largest range of contributing participants.Question 14 was noted to have the highest sequential interaction, with 163 interactions, but only 28 engaged participants contributed to the sequential interaction, which was relatively low when compared with other questions.This means that the contributing participant range was low.Therefore, we developed two quasi-statistic measures which we viewed as more meaningful.In combining these (sequential interaction and the number of contributing participants), we identified the highest and lowest sequential interaction questions, the combination of which helped us to determine the questions that created the most impact in this research.
Sequential interaction measures (i.e.quasi-statistics) helped identify how impactful questions were (in our case, four were high and two low impact).We also noted that not one of the lowimpact questions generated any of the identified interactive features of laughs and pauses.
Our results indicated that both groups A and B identified Q10, 'What are the various ways the community can benefit from giving?-Can you give examples?',as the question that had the highest number both of sequential interactions and of contributing participants.We therefore classed Q10 as the question that created the highest levels of impact in this research [Table 3].We believe this was because those attending the focus groups had strong views about corporate philanthropy and who should benefit from that giving.Q9 was planned as a higher-order question; 'How does John Lewis make decisions on which charities/community groups get support (when there are so many competing for help)?'However, surprisingly, different groups had different reactions to it.We believe that a potential reason for the difference could be that a new organizational giving initiative was initially piloted in the Edinburgh store.

Discussion
Earlier findings and previous insights agree that sequential interaction analysis and the presentation of results is an unappreciated area of focus group research.Rinkus et al. (2021), Cyr (2016), Duggleby (2005), Halkier (2010), and Kitzinger (1994) all concur that research papers often fail to expand on how to capture and analyze interaction, with reporting of the conversations between participants rare.Wibeck et al. (2007) consider interaction to be a hallmark of focus groups, but state that it is seldom evaluated, often ignored in analysis, and rarely discussed in empirical research.We argue that researchers should go beyond merely extracting and applying quotes to participants as they might with an interview.
Responding to Morgan's (2010) ambitious three-stage agenda for improving our understanding of the role that interaction plays, we have provided a detailed, reflexive investigation into how interaction could be conducted through a method of capturing and evidencing sequential interaction.By analyzing interaction, researchers can distinguish between the responses of different focus groups, make connections between a taxonomy of questions and the associated contributors, and identify those areas of most significance within group discussions (impact).
Our purpose was to develop a 'framework' through a two-stage approach, designed to capture and evidence sequential interaction.The aim was to advance an approach that is easy to follow, and replicable yet also adaptable.The reporting of quasi-statistics (Barton, 1995;Becker, 1970;Lazarsfeld, 1935;Warr, 2005) allows for comparison between focus groups and offers a more complete approach to analyzing interaction results (incorporating what we term 'sequential interaction').Using the approach described, researchers and practitioners alike will be able to encourage and generate more interaction between participants and also record additional features of interaction information (referred to as quasi-statistics), such as pauses, tears, and laughter.Following the two-stage approach identified here, researchers will be guided to locate the particular questions that create the highest interaction impact in focus group research.To achieve this, we have employed quasi-statistics by both focus group and question, supported by a transparent audit trail designed to capture and evidence sequential interaction.
There is consensus that the reporting of sequential interaction is a key dilemma in focus group research.One of the core questions is why researchers start analyzing at Q1, then work methodologically 'question by question' as if no sequential interaction has taken place.By highlighting different interactions and responses within groups and examining these results, we can explore reasons why differences occur between groups and what meaning this may have for our and future studies in the area.
In order to demonstrate our framework, we carried out a study of corporate philanthropy working as outsider researchers with the organization John Lewis & Partners.In our research we reduced a series of eight focus groups containing 64,572 words of transcription into 2230 sequential interactions with specific high-impact questions.The identification of high-impact questions can be used as a starting point for new findings and insights, developing new hypotheses and subsequent theorizing (Bratton & Liatto-Katundu, 1994).Our study also identified issues of practical relevance, which we discuss in the next paragraph.
Practitioners can appreciate deeper insights into meanings and explore areas of difference and consensus through the signposting of high-impact questions.For instance, differences in the identification of the high-impact question across a series of focus groups leads to insightful knowledge.In our study, the results highlighted some interesting differences.Groups A and B identified Q10 as the question that had the highest number both of sequential interactions and of contributing participants.In contrast, in Q9, in different locations produced different levels of sequential interaction.This provides practical relevance for the organization to explore reasons for difference.The framework leverages and promotes the use of focus groups while adding rigor to the method, revealing a new understanding to identify questions of impact and leads us to an increased point of knowledge for the researcher to commence qualitative theorizing.We believe those questions that generate most discussion within a focus group (and hence interaction) are among the most important when analysing the results of a discussion (however, we do not claim that they are the only important questions).This helps researchers in identifying the key issues which arise during group meetings.It could be used for piloting questions for future groups, or it could be used to identify issues which are most important to employees.For example, it could be used to identify which charities are most important to employees of specific branches of a company and thus used to determine corporate philanthropy policy within an organisation.Comparisons could also be made between the viewpoints of different offices or branches of an organisation as determined by their respective focus groups and their interaction with different questions.
Future research is required to evidence, report, and present focus group interaction within different organizational settings.There may also be other ways of considering and documenting sequential interaction.(1981) indicates that all studies inevitably contain some flaws.The aim of our paper is to encourage interaction between participants because this is where we believe key issues are identified within focus groups.To this end, we have suggested various interventions such as careful design of questions, the moderator adopting a facilitator/ learner perspective and careful planning including the consideration of room layout.However, despite these interventions there remain other variables than may determine the amount of interaction within a focus group, for example, the way a group gels together, the existence of dominant or passive individuals within groups.Not all variables can be planned, anticipated, or controlled for.However, we are interested in generating discussion in the literature so as to further address these limitations in such a way to maximize group interaction.

McGrath
Focus groups often do not subscribe to ideal settings.In our study, a particular issue was evidenced in focus group F, where participants were talking over each other in an animated way, including some using a strong dialect.This made it difficult to work out what was being said and also to record quasi-statistics.The number of contributing participants was not tracked and was therefore discounted from the quasi-statistics.This was an important issue for those contributing and clearly contributed to a stimulating or emotive question, which resulted in the lively sequential interaction between participants which we observed but could not adequately document.
Sequential interaction leads us to the identification of the questions that promoted the highestand lowest-impact conversations, alongside other features of interaction such as laughs and pauses.A dominant voice or lengthy sequential conversation between only a small number of participants was also of concern.We attempted to address this by introducing the range of contributing participants' measure as a quasi-statistic.
The number of researchers as moderators dedicated to conducting focus groups is also a limitation.One moderator is not sufficient to set the conditions for interacting, observing, and capturing non-verbal interactions, while simultaneously asking questions.Using two moderators is a possible solution to this.
In analyzing this sort of data, it is useful to consider using software (e.g.NVivo) as part of the process.We chose not to not use this technology, remaining as close to the data as possible.Introducing technology as a data analysis tool for sequential interaction analysis may have a cost implication and place the researcher at a distance from the data, but it can also be time efficient.

Conclusions and Future Research
It is important in focus group work to consider exactly how to encourage, capture, evidence, and present sequential interaction between the various participants.Without an attempt at investigating interaction, analysis in such research lacks completeness and rigor, missing out on useful data.Morgan (2010) affirms that failure to quote interaction in focus groups devalues the interaction process that produced the data.Different questions and different groups may give rise to different levels of interaction, so any approach to focus group interaction needs to take this into account.
This paper has been devoted to developing an exploratory two-stage framework to capture and evidence interaction in focus groups, creating new discourse previously not entertained.We have detailed how we applied our 'framework' to our own particular study of philanthropy.The extent to which the researcher operationalizes and evidences sequential interaction analysis into achievable iterations and steps is inevitably subjective; there is little precedent regarding choices and analysis.However, there are three main stages to be undertaken prior to the commencement of the qualitative analysis: (1) The initial stage is to design the questions in such a way as to maximize interaction.
(2) The second stage involves the collecting of quasi-statistics which can help identify areas of impact.The researcher (and practitioner) can decide exactly what quasi-statistics they capture: not all will be useful, and this area may involve an element of trial and error.(3) Finally, our research has suggested ways of monitoring impact based around the amount of sequential interaction.
Whilst we want to underplay the content/qualitative understanding the significance is attributed to the disclosure of nuances between focus groups, establishing a point of increased knowledge for the theorizing approach.We acknowledge that, although interaction is a relatively simple concept, it is quite difficult to capture.

Future Research
This could take the form of a number of different avenues.First, suggestions for altering or adding to the framework (or even devising alternative frameworks) would be useful.Second, it would be useful if different researchers tested the framework in different contexts.One advantage is that the framework can always be adapted to the circumstances.In our study we did not expect participants to become so animated.The norms of not speaking over each other were compromised, and, while this only occurred in one group, if it became a common occurrence, it would have been fairly easy to add to the framework and as a result help identify questions of particular impact.

Notes
1. Additional tables are available from the first named author.2. Although the quotes used have been edited for grammatical errors, the meaning remains unaltered.

Disclosure Statement
No potential conflict of interest was reported by the author(s).

Notes on contributors
Helen Nicholson is an Assistant Professor at Northumbria University.Helen made a mid-career change into academia from the hospitality industry.This move provided an appreciation of practical relevance and theory.Her research interests lie within employee-owned business, exploring shared governance, inclusive decision-making, and the advancement of focus group research.Helen lectures in marketing, research methods for DBA/PhD candidates and is a practitioner of experiential learning.She has recently published in the Journal of Business Ethics.

Table 1 .
Quasi Statistics by Focus Group.Stage Two -Iteration One -Quasi Statistics by Focus Group.

Table 2 .
Quasi Statistics by Question.Stage Two -Iteration Two -Quasi Statistics by Question.

Table 3 .
Identification of Impact Questions.Identification of Impact Questions.
Shrives is Emeritus Professor of Accounting and Corporate Governance at Northumbria University, UK.He is a Chartered Accountant and a Fellow of the UK Higher Education Academy.Prior to his retirement he lectured in accounting, risk reporting, corporate governance, impression management, and research methods.He was also previously Director of Academic Staff Development.Philip has authored articles in a number of prestigious journals including The British Accounting Review, Critical Perspectives on Accounting, and Accounting, Auditing and Accountability Journal.