• No results found

CHAPTER 5: METHODOLOGY

5.3 Data generation

5.3.4 Data generation methods

For the study, I planned to employ the following as data generation methods: 1) individual interviews with young refugees and key informants 2) observation of pages and groups on social media aimed at refugees in Norway, and 3) an online survey of interview participants‘

daily use of social media. The first two methods were used, but the last was discarded.

1) Individual semi-structured interviews with young refugees were carried out using an interview guide with open-ended questions (see Appendix 2). All interviews were conducted face-to-face, in English, by me. Participants were contacted prior to the interviews by text or email, when a brief outline of the project was provided to them. Some participants also saw copies of the flyer used to advertise for research participants (see Appendix 1). At the start of each interview, the purpose of the project was explained in more detail.

Locations for the individual interviews were guided by the interview participants. Most participants suggested a location that was convenient and comfortable for them. Three of the interviews took place in cafes, one in a shopping mall, one in a hotel, two in a library, and one in the interviewee‘s own home. At each interview there was only myself and the interviewee present. All settings offered enough privacy for participants to speak freely and to enable audio recording of the interview.

34 The interviews ranged between 25:00 minutes and 45:00 minutes in length. All participants agreed to have their interview audio taped, using a digital voice recorder, which I later transcribed. The interviews were recorded in order to ensure a truthful recording of what was said, provide accurate data for analysis, and to enable me to be fully present and focussed during the interview. The interview guide was developed to incorporate key factors relevant to the study, including 1) uses and gratifications of social media use, 2) outcomes or

capabilities enabled by social media use, 3) messages and campaigns aimed at refugees on social media. The interview guide included a definition of social media which was explained to participants prior to the start of the interview. A printed ‗prompt sheet‘ containing 22 popular social media icons was also shown to participants, so that they could refer to it during the interview as a reminder of the wide range of social media sites available (see Appendix 3).

Interviews were also conducted with two key informants – one interview, with a moderator of the ‗Refugees Welcome to Bergen‘ Facebook group, was conducted face-to-face; due to time constraints and practicality, an email interview was conducted with a representative of the Norwegian Ministry of Justice and Public Security, which runs the ‗Stricter Asylum Regulations in Norway‘ Facebook page. Key informant interviews were not originally planned as part of the methods for this study, but due to the failure of another data generation method (the online survey), I decided to conduct these late in the data gathering process. As Skovdal and Cornish (2015, p. 56) point out, key informant interviews provide quick access to important facts, from carefully selected individuals who have access to those facts and can be used to inform a rapid appraisal of a situation. An interview guide (Appendix 2) was used for the face-to-face interview. Questions were prepared for the email exchange, which were emailed to the informant to answer.

2) Online observation was conducted with two Facebook groups, ‗Refugees Welcome to Norway‘ and ‗Refugees NOT Welcome to Norway‘, and one Twitter feed, UDI‘s

@Utlendingsdir. The rationale for choosing these three sites was to observe one ‗pro‘ and one

‗anti‘ refugee group, and one site run by the Norwegian government. Whilst the ‗Stricter Asylum Regulations in Norway‘ Facebook page was initially selected as the government site to observe, it was not active enough (the most recent post to the group was posted three months prior to the start of the observation period). All sites had to be ‗active‘, with

up-to-35 date ‗live‘ posts and interactions, since old posts would be classed as historic material. Due to ethics considerations, it was also only possible to observe ‗open‘ or public Facebook groups, which significantly limited choice. I initially planned to observe the ‗Refugees Welcome to Bergen‘ group, but since this was a closed group the public ‗Refugees Welcome to Norway‘

group was chosen instead. (A key informant interview was later conducted with a moderator of the ‗Refugees Welcome to Bergen‘ group). Moderators of both the Facebook groups chosen were contacted and their permission was sought and obtained to observe the groups prior to the start of the observation.

The observation period took place between 5 September 2017 and 3 October 2017, with the three online sites observed twice a week each. Participant observation is often conducted on a short time frame (from a few weeks to a few months) and overlaps with other forms of data collection (Guest, Namey & Mitchell, 2013, p. 100). Observation field notes were taken, using an observation guide adapted from Skovdal and Cornish (2015, p. 161) (see Appendix 4). I chose to make field notes following guidance from the literature that they are necessary even in online settings and that it is essential for observers to record their comments,

including reactions and interpretations (Salmons, 2016, p.123).

3) An online survey for interview participants was created using SurveyMonkey, a free online survey tool often used in academic research. The rationale for the survey was to provide a

‗snapshot‘ of the realities of how participants were using social media in their everyday lives, to supplement the information that they gave in the interviews. The survey link was given to each interview participant and they were asked to complete it on a daily basis for a period of one week after the interview. The survey was designed to be deliberately short, with only four questions, so that it would not be cumbersome or time-consuming to complete (see Appendix 5). It was also set up to be anonymous. IP addresses would not be collected with survey results and there was no login or identifiers required to access it.

However, although the survey was set up and links were given to all interview participants, by the end of the agreed data generation period it had not completed by any of them. This method of data generation was subsequently discarded. The failure was likely influenced by the fact that the survey links could not be sent to participants electronically (in order to ensure anonymity and comply with National Centre for Data Research, NSD, requirements), as several participants requested, but instead had to be handed to them on a piece of paper. It

36 was likely that this would have been lost or forgotten. Given the failure of this method, I decided to instead conduct key informant interviews with individuals who had insights into the phenomenon of social media use with refugees in Norway as another method of data generation.