Causal inference and survival analysis
Torben Martinussen, Department of Biostatistics, University of Copenhagen
Causal inference techniques has become very important tools in modern data analysis.
There has been much debate whether special issues arises when one is considering
time-to-event data. Classical tools for such data have evolved around the hazard
function with the Cox regression model being the primary example. However, concerns
has been raised whether this is a suitable function to consider when trying to draw
causal inference, and one even speaks of "the hazards of hazards". In this talk I
will try to shed some light on these issues.