Visualising Q - from the bootstrap to interactive graphics

Abstract Introduction The MIDAS (Meaningful Integration of Data Analytics and Services) project is developing a big data platform to maximize the use of health & social care data. The goal is to link data sources to support senior managers & policy makers in delivering services. The project includes 4 case studies, in 4 countries. As part of the evaluation, we need to understand the perspectives of the users, developers, and senior managers involved, and to see how these change over time. We employ Q-methodology, an objective mixed method for the study of human perspectives, to do this. In this paper we describe the use of bootstrap methods, and visualizations to assist in the execution and interpretation of the first round of our Q study. Methods A concourse of 36 items was developed from the literature, a logic model for the project, and a series of semi-structured interviews with project participants. Sixteen people (3 female, and 13 male) took part in the Q study, six developers, five managers, two health professionals, and 3 others. The 36 statements on the concourse were ranked online, by each participant, using the HTMLQ software, in order of their agreement with each statement. These are then subjected to a form of factor analysis, but by person, not by statement, using the qmethod package in R. For each Q-sort 1,000 bootstrap replications were done, using sampling with replacement. A range of visualisations were prepared, using ggplot2. Results Visualizations of bias and variability showed modest levels of both, suggesting that the Q-method model fitted well. Interactive visualizations of the factors, and respondents, were done. These showed distinct clusters of respondents, with divergent perspectives on the project. These assisted in making final decisions, both on the number of factors to report on, and the interpretation of those factors. Further use of advance visualisations is recommended for future Q-studies. Funded by the European commission under contract 727721 Key messages Q methodology is useful across many areas of public health, and is a valuable way of studying individual perspectives. Modern statistical visualisation tools can enhance the interpretation of Q methodology studies.