Skip to main content

Table 2 Four characteristics (with examples) of successful Collaborative Data Analysis

From: Best practice framework for Patient and Public Involvement (PPI) in collaborative data analysis of qualitative mental health research: methodology development and refinement

1. The CDA process is co-produced
 • Keep consulting; verify everything with PPI co-researchers [48]
 • Good facilitation with a supportive and valuing approach is essential. Understand the perspectives/positions of the PPI co-researchers interpreting the data. Be reflexive Be mindful of the personal investments people can hold in how topics are interpreted. [31]
 • Support PPI co-researchers to understand that while experience can be used to help interpret the data, all interpretations must have some basis in that data [50]
 • Ensure a range of perspectives amongst the PPI co-researchers interpreting the data; aim for a heterogeneous group [31, 50]
 • Listen to and explore differences of opinion. When non-consensus occurs, try to create novel synthesis to acknowledge the range of perspectives [31]
2. The CDA process is realistic within available time and resources
 • Ensure sufficient resources exist, e.g. time and money to organise and facilitate CDA. Do not underestimate this [31]
 • Keep the number of analysts relatively small [30]. Use software packages to investigate inter-coder reliability if not everyone is coding all data [31]
 • Make sure data handling and organisation is meticulous [31]
3. The demands of the CDA process are manageable for PPI co-researchers
 • Give PPI co-researchers material to read in advance [48, 49]. Make sure materials are accessible and in a range of formats where required [49, 50]
 • Provide training that ensures people can successfully complete the CDA they have been asked to do [49, 50]. Do ‘warm up’ activities that align with the CDA tasks people are being asked to undertake [48]. Use practical, visual aids like post-its and flip chart paper to support analysis tasks [48, 49]
 • Keep the data set relatively small and do not present people with too much raw data [30, 48]. Ensure the data analysis process is adjusted to take into account the strengths and needs of PPI co-researchers and is ‘failure free’ [32]. Allow ample time for analysis [30]
4. Group expectations and dynamics are effectively managed
 • Clearly set out the PPI co-researcher role and expected time commitment [31, 49, 50], and how their contributions will be valued and incorporated [31]. Clarify the division of labour (in writing if appropriate) [31, 49]
 • Be mindful of labelling: people hold multiple identities and categorisation can cause inter-group tensions. Be vigilant for power imbalances, which may occur even with the best of intentions [31]