Pearl and Mackenzie [2018] provide a readable overview of causality and its historical context. For a more technical description, see Pearl [2009], Spirtes et al. [2001] and Veitch and D’Amour [2023]. Geffner et al. [2022] overview the work of Judea Pearl, one of the pioneers of causality in AI, and include many papers on causality.

Huang and Valtorta [2006] and Shpitser and Pearl [2008] independently showed that the do-calculus is complete; it completely characterizes when interventions impact probabilities given only the causal structure.

Modeling missing data is discussed by Rubin [1976], Little and Rubin [1987], Marlin et al. [2011], and Mohan et al. [2013]. Mohan [2022] provides an overview.