>[Birnbaum's] solution is twofold: firstly he ‘proved’ LP from two premises which any frequentist is supposed to accept (Birnbaum 1962), but secondly nevertheless abandoned LP for the sake of fruitfulness brought by sampling theories (Birnbaum 1969)
https://t.co/O3sfkb5UpM
Our new pedagogic paper with a fresh numerical example for longitudinal causal inference is published as a part of “Pitfalls and Tips” series in Journal of Epidemiology
Understanding marginal structural models for time-varying exposures: pitfalls and tips https://t.co/UNX3L7l5VP