Translational & Implementation Science
Expert knowledge from living experience: Applications to Directed Acyclic Graphs Megan Marziali* Megan Marziali Peggy Frank Kathleen Inglis Wayne Campbell Sandy Lambert Patience Magagula Silvia Guillemi Catherine Worthington Robert Hogg Valerie Nicholson Michael Budu Melanie Murray
Aim: Causal inference is not possible without background knowledge and relies on the background knowledge being correct, as it is encoded in identifiability assumptions. It is generally understood this knowledge comes from academic subject matter experts; people with living experience should also be considered experts, providing unique and novel insights to causal questions. Our study blends living experience with scientific domain knowledge to construct the directed acyclic graph (DAG) underpinning our research question about aging and HIV, challenging the notion of what constitutes “expert” knowledge.
Methods: We assembled a team of quantitative and qualitative, including diverse peer researchers (i.e., People Living with HIV). To train peer researchers and other team members unfamiliar with DAGs, we organized a retreat where peer researchers watched short tutorials on building a DAG. Quantitative research team members created these tutorials to define types of bias (e.g., confounding, selection bias) and outline DAG rules. The team worked together to create a DAG illustrated by a peer researcher-artist, combining the living experience of peer researchers with scientific and medical knowledge from quantitative researchers. The DAG was then refined with input from epidemiologists. Qualitative researchers transcribed the process of learning about and drawing the DAG to understand how peer researchers engaged with this tool.
Results: Peer researchers found drawing DAGs intuitive. Visualizing the temporal ordering of variables was a crucial moment in understanding the causal structure for peer researchers. The complexity of the DAG enmeshed with temporal ordering humanized the research; seeing the web of covariates resonated as a reflection of a person’s life journey.
Conclusion: Constructing DAGs incorporating living experience is feasible, ethical, and strengthens epidemiologic studies by reconceptualizing our definition of background knowledge and humanizing research.