2.6. Data Analysis
Survey data were analysed using IBM PASW SPSS Version 25.0 (IBM, Armonk, NY, USA) (Supplementary File S6). Data cleaning procedures (e.g., identification outliers and missing data analysis) and key statistical assumptions underpinning t-tests, correlation, and linear regression (normality, linearity, homoscedasticity, and independence) were examined prior to data analysis. Qualitative data from interviews, focus groups and open-ended survey questions were analysed using inductive thematic analysis, which benefits from theoretical flexibility and simplicity in the identification of qualitative themes [13]. This process included the in-depth familiarisation and coding of data using NVivo 12 software, before sorting data in broader thematic concepts which represented sections of the data, later refined into the development of five key themes, and 13 subthemes. Two researchers (LB/CC) analysed qualitative data, using thematic analysis [13]. As this was an evaluation of a pilot uncontrolled complex intervention in a real-world setting, intended to directly inform ensuing mass SARS-Cov-2 testing approaches, a pragmatic and time-sensitive approach was taken to analysis. Three researchers were involved in the qualitative analysis (LB, CC, HB). One researcher coded all the interview data and generated the initial themes (LB), a second researcher (who had conducted interviews) then independently coded a subsample of four randomly selected transcripts, in order to compare and agree on themes through discussion. A third researcher (who had conducted interviews) then reviewed all the transcriptions to crosscheck against themes, confirm the themes and resolve any discrepancies between coders (HB). Consensus on the themes was achieved through discussion between all researchers. Combining qualitative data from different data sources and using two researchers for coding and analysis, enabled data and investigator triangulation.