We conducted a sub-study within a study that was examining the well-being of older people on antiretroviral therapy (HIV treatment) in a rural setting in Uganda to explore the differences between data produced from audio recordings and those produced from notes and memory of the interview. In this study – which was led by one of the authors of this paper, JM – data were collected by two trained and experienced interviewers (both with more than 20 years of experience of qualitative data collection) (co-authors EK and GT) who used voice recorders as a data capture method. They conducted 30 interviews each. All interviews were conducted in Luganda, the main local language. As part of our sub-study, the interviewers, who were also trained and experienced in data capture without the use of voice recorders, were asked to prepare scripts in English of some of the interviews they had conducted without listening to the voice recordings. Preparation of these scripts was done immediately after each interview, preferably the same day, to minimise recall bias. Scheduling time for this activity was important for data management. The voice-recorded interviews were transcribed verbatim and translated into English by someone other than the interviewer. For this paper, 60 pairs of scripts/transcripts (each pair consisting of the transcript from the recorded interview and the script from the notes of the interview) were analysed and compared. The initial comparison was conducted by EK and GT. Subsequently four other co-authors of this paper (RR, JM, MM and JS) conducted further analysis and cross-checking to verify the initial findings.
Free full text: Click here
