Are all retrospective studies exploratory?

hlsmith

Omega Contributor
#2
No.

Those that say no causation without manipulation may disagree.

But many strive to get as close as possible to RCT data using observational data. The focus needs to be on examining Bradford Hill criteria along with understanding treatment assignment mechanism and establishing covariate balance between groups using statistic approaches.
 
#3
I think I got Your point.

If we decide to test if particular factor affects retrospectively acquired survival data, we have just created new concrete hypothesis that is not exploratory because this factor was not "fished for" among many other prerecorded factors (instead it was obtained by apriori reasoning)?
 
Last edited:

hlsmith

Omega Contributor
#4
A other example is the interrupted time series or regression discontinuity using historic data. It is not exploratory and has components that get it close to discerning causality.