Improving the use of focus group discussions in low income settings

BACKGROUND: The quality of data obtained through Focus Group Discussions (FGDs) is highly dependent on appropriate design and facilitation. In low-income settings steep power gradients between researcher and participants, as well as conversational norms, could reduce the ability of participants to voice personal opinions. Activity-oriented exercises have been suggested as a way overcoming these challenges, however little evidence exists - to date - on their use in low-income settings. We selected six exercises for use in Ethiopia and Nigeria and report our experiences. METHODS: The six exercises (picture sorting, associative pictures, picture ranking, decision trees, predictive story-telling and provocative statements) were used in 32 maternal and new-born care themed FGDs conducted in Amhara and Southern Nations Nationalities and People’s Regions (Ethiopia) and Gombe State (Nigeria). Six facilitators and two supervisors who used these exercises were interviewed about their experiences. FGD verbatim transcripts and interview notes were analysed to explore methodological effectiveness and respondents’ experience. All data were coded in NVIVO using a deductive coding frame. RESULTS: Facilitators and participants described the methods as ‘fun’ and ‘enjoyable’. The exercises yielded more in-depth and complete information than ‘normal’ FGDs, but facilitator’s probing skills and overall FGD group dynamics proved crucial in this success. Explaining and conducting the exercises increased FGD length. Data richness, participant reaction and understanding, and ease of facilitation varied by study site, exercise, and participant group. Overall, the exercises worked better in Nigeria than in Ethiopia. The provocative statement exercise was most difficult for participants to understand, the decision-tree most difficult to facilitate and the picture exercises most enjoyable. The story telling exercise took relatively little time, was well understood, yielded rich data and reduced social desirability bias. DISCUSSION: The majority of the exercises proved successful tools in yielding richer and less biased information from FGDs and were experienced as fun and engaging. Tailoring of the exercises, as well as thorough training and selection of the facilitators, were pivotal in this success. The difference in the two countries shows that adequate piloting and adaptation is crucial, and that some exercises may not be adaptable to all settings. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12874-020-01168-8.