New blog post up: using causal missingness DAGs to understand missing data and choose an appropriate analysis. Ever been confused by terms like MCAR, MAR, MNAR? Wondered if there's any way to deal with data being Missing Not At Random? This might help. cameronpatrick.com/post/2023/06/u…

Possibly dumb question for longitudinal data analysis people: Why is it necessary to condition on subject in a repeated measures design? (For instance, with a random intercept). If your model includes a term for time, aren't observations then conditionally independent?

Ever wondered what "noncollapsibility" of odds ratios meant? Not sure what an odds ratio is, but too afraid to ask? Wondering how best to report effect sizes for binary outcomes? I accidentally wrote almost 4,000 words on the topic... cameronpatrick.com/post/2023/07/l…

👀Development of the TARGET reporting guideline👀 We've assembled an all-star team to develop a #reportingguideline for observational studies explicitly aiming to emulate a target trial. Reporting of these studies is variable #epitwitter #causalinference doi.org/10.1136/bmjope…