Focus Groups

Master Thesis, May 2018 Olivia Paludan Wegmann Copenhagen Business School

34

The Danish YouTuber, Simone Wulff, with 113,400 subscribers on her channel accepted the interview invitation. (Unfortunately, the other YouTubers replied that they do not participate in school-related interviews). The interview with Simone was conducted over the phone, something Struges and Hanrahan (2004) find especially relevant when trying to access more hard-to-reach respondent groups. An interview guide (Appendix 3) was developed, and a semi-structured approach was applied. This presents a flexible form of interviewing (Kvale, 2011) and enables the researcher to ‘stir things up’ at any point in the process by introducing spontaneous follow-up questions (Kozinets, 2010). After the interview, the data were transcribed from audio records into written form (Appendix 4). The data were analysed through colour-coding, and the colours and related themes are apparent from the transcript (Appendix 4).

Master Thesis, May 2018 Olivia Paludan Wegmann Copenhagen Business School

35 4.4.2.1. The Focus Group Instruments

According to Stewart et al. (2007), three research instruments influence the success of a focus group interview: (1) the interview guide, (2) the moderator, and (3) the group itself.

4.4.2.1.1. The Interview Guide

The interview guide sets the agenda for the group discussion and has the purpose of providing a direction for the discussion (Stewart et al., 2007). The questions in the interview guide (Appendix 5) grow directly from the problem statement and sub-questions, and the goal is to obtain data that can answer them.

Stewart et al. (2007) identifies two general principles that should be taken into account when formulating the interview guide: (1) “the questions should be ordered from the more general to the more specific”, and (2) “questions should be ordered by the relative importance to the research agenda” (Stewart et al., 2007, p. 63). The focus group interview guide for this study follows these guidelines, as questions of the highest importance are kept more general and unstructured, compared to the less important questions, placed towards the end.

The agenda for the interviews was semi-structured, meaning a set of questions were prepared, yet it was possible for new questions to emerge during the conversation. Also, the structure of the questions was fluid, allowing the moderator to ask questions earlier than anticipated, if appropriate for the discussion.

A combination of open questions and more specific questions was included. Open questions allowed the respondents to draw attention to any dimension of the general stimulus identified in the question, and specific questions enabled respondents to reflect on a particular aspect of the stimulus object (Stewart et al., 2007).

The interview guide was pre-tested prior to the focus group interviews by asking the questions to a respondent with the same characteristics as the focus group sample. This pre-test had the purpose of determining the appropriateness of the wording of the questions, whether the questions raised a discussion, and whether the questions were easily understandable.

It is important to recognise that the groups may take on a life of their own, meaning the interview guide should function as a guide which the moderator and group can modify, if need be (Stewart et al., 2007).

4.4.2.1.2. The Moderator

According to Stewart et al. (2007), an effective moderator is “one of the keys to the collection of rich and valid insights from focus groups” (p. 69). Krueger (1997) lists three types of moderators where the moderator of the focus groups in this thesis can be categorised as a volunteer, as it is

Master Thesis, May 2018 Olivia Paludan Wegmann Copenhagen Business School

36

the researcher herself who moderated the interviews. The advantage of this choice is the moderator has “adequate knowledge of the topic”, “appears like the participants”, and can

“exercise mild unobtrusive control” (Krueger, 2002, p. 2). Stewart et al. (2007) find that the more compatible the interview participants and the moderator are, “the greater the interaction and the more open communication” (p. 81). Meanwhile, the disadvantage is the moderator’s limited experience with focus groups. Where a professional moderator would have more experience to draw upon, a volunteer has “a unique credibility and insights on the topic” (Krueger, 1997, p. 37).

Furthermore, it is possible to distinguish between the different roles of a moderator. In this study, the key role is to be the seeker of wisdom, meaning the goal is to “obtain understanding, insight, and wisdom”, whereby it is assumed that the respondents have such wisdom, as well as the willingness to share it (Krueger, 1997, p. 46).

4.4.2.1.3. Group Composition

In general terms, the composition of the group is of high importance as the quality of the discussion is likely to be determined by the interaction between the people who are brought together (Stewart et al., 2007).

Two relatively homogenic groups of people were recruited in accordance with the specific character of the research problem. The purpose of two focus groups is not a comparative analysis, but to gather as much knowledge about the problem as possible. The homogeneity of the group is an element that may influence the dynamics, and this will be covered in the data analysis.

The following characteristics were required from the participants: They had to be female and 13-16 years old, regular viewers of beauty and fashion on YouTube, and live within driving distance of the interview location in Copenhagen, Denmark. The participants were recruited through the help of the researcher’s network. Three friends-of-friends were asked to identify relevant participants who were invited to join the interviews. Thereby, the interviewees were not familiar with the researcher (and moderator).

Meanwhile, some of the participants were acquaintances of each other, posing implications that are necessary to be aware of. When some participants know each other, it is possible that they will not fully articulate assumptions and contexts, as they already possess knowledge about each other (Stewart et al., 2007). Therefore, the moderator must assess whether and when some answers need elaboration. Additionally, such a group composition will most likely generate different data, than if they were produced in a group of strangers, or even from the same participants, interviewed in a different context (Stewart et al., 2007).

Focus Group 1 consisted of four girls at the age of 14-15, and the interview was facilitated on March 17, 2018. Focus Group 2 was held on March 25, 2018 and consisted of seven girls at the

Master Thesis, May 2018 Olivia Paludan Wegmann Copenhagen Business School

37

age of 14. The interviews took place at a private home in Copenhagen, Denmark. Conducting the interviews in a private home may make the participants feel more at ease and in balance. Cake and snacks were served, as the presence of food may help relax participants (Stewart et al., 2007).

Also, according to Stewart et al. (2007), the seating arrangement influences the group members’

degree of participation, leadership behaviour, and patterns of interaction. Therefore, the seating allowed all members to see one another to reduce any tendency of some of the participants to be dominating, and avoid the emergence of subgroups (Stewart et al., 2007).

4.4.2.2. Qualitative Data Analysis

4.4.2.2.1. Transcription

The starting point of the focus group data analysis was transcriptions of the interviews (Appendix 6;

Appendix 7). The sessions were transcribed from an audio recording into written form in full length (ad verbatim), meaning that elements such as incomplete sentences or half-finished thoughts were included in the transcripts. The interviews were held in Danish, and the transcripts were kept in Danish as well. When quoting the participants in the data analysis, the quotes are translated into English as directly as possible to avoid any loss of meaning or accuracy.

4.4.2.2.2. Coding of the Transcription

The ‘scissor-and-sort technique’ was applied to analyse the transcripts (Stewart et al., 2007), and this involved two steps: (1) Reading the transcripts and identifying sections of relevance to the problem statement, and thereafter, (2) formulating a classification system for the major topics in discussion. This classification system centred on the uses and gratifications expressed during the conversations, and the following five themes were colour-coded: Utility, gratifications, perceived influence, interaction and dialogue, intermediality and competing media. The colour coding is apparent in the enclosed transcripts.

The approach to code the transcript is useful and efficient, yet it relies on the judgment of the researcher. So, the findings may be biased by what the researcher subjectively finds important.

Meanwhile, Stewart et al. (2007) notes that the approach shares “many of the characteristics of more sophisticated and time-consuming approaches” (p. 124). In an ideal scenario, two or more analysts would have coded the transcripts.

Master Thesis, May 2018 Olivia Paludan Wegmann Copenhagen Business School

38