upcoming holidays or starting term, because they allow the interviewee to return to the real world.

Having recruited the interviewees and developed the guide, I started conducting the interviews in December 2018 in the study rooms at the LSE Library due to its easy accessibility and quietness. Reflecting on the first interview, I could have let the content drive the process a bit more as opposed to forcing the structure on the interview said by my questions. The learning from the first interview made the second one more relaxed and organised at the same time.

However, I transcribed the interviews only a week later and not at the earliest opportunity. Luker (2008) argues the necessity of immediate transcription because reading written transcripts work differently for the mind than just listening back the records, and early transcription leaves a higher chance of capturing key takeaways. With hindsight, I would transcribe the interviews immediately because, while transcribing the first two interviews, I grasped that there were many spaces to improve both my question and my interviewing style.

Reflecting on my interviewing style in the first interviews, my interactions could have been more colourful. For example, I replied with an “Okay” or “Good” to many of the responses. Occasionally I also interrupted the interviewees with those reassuring reactions, whereas I should have listened silently. Likewise, due to my insider position in the researched topic, I talked more about my own experience than I should have. This behaviour goes against what Weiss (1994) claims about self-introduction of the researcher, according to which it should be decreased as much it is consistent with the interview. Additionally, I wanted to control the interview too much with continuously asking questions, thus not giving time to the interviewee to elaborate more on some topics. This could be considered to be an improper way to conduct an interview since an interview is regarded to be a collaboration of the interviewer and interviewee (Weiss, 1994). The transcription also shed light on which questions worked and which were not that successful. These insights benefit that the researcher gains from the first few interviews (Luker, 2008). For example, it turned out that questions regarding family contacting habits proved to be too general.

To address the weaknesses of the first interviews, I worked on the improvement of both questions and interviewing style. To tackle the dilemma of too general questions and the resulting obscure answers on the central topic, I followed Weiss’s (1994) guidance on how to obtain more detailed responses. In the topic of family contacting habits, I focused more on asking for the explanation of specific incidents, like “Could you walk me through the